Optimisation of the Reaction Parameters in a Batch Reactor and a CSTR for the Recovery of Phenol from Hydrothermal Biomass Liquefaction

Optimisation of the reaction parameters in a batch reactor and a CSTR for the recovery of phenol from hydrothermal biomass liquefaction

zur Erlangung des akademischen Grades eines DOKTORS DER INGENIEURWISSENSCHAFTEN (Dr.-Ing.)

der Fakultät für Chemieingenieurwesen und Verfahrenstechnik des Karlsruher Institut für Technologie (KIT)

genehmigte DISSERTATION

von Dipl.-Ing. Daniel Forchheim1,2 aus Karlsruhe

Referent: Prof.Dr. Michael Türk1 Korreferentin: Prof.Dr. Andrea Kruse2 Tag der mündlichen Prüfung: 25.03.2014 1 Institut für Technische Thermodynamik und Kältetechnik 2 Institut für Katalyseforschung und -technologie Acknowledgements

An dieser Stelle möchte ich besonders meinen Eltern danken, die mich mein Leben lang unterstützt haben, der Forschung und den Wissenschaften nachzugehen. Ich danke Andrea Kruse und Uschi Hornung, die mir mit vielen Hinweisen und in zahlreichen Gesprächen wichtige fachliche Unter- stützung und Vertrauen in meine Arbeit entgegengebracht haben. Ein beson- derer Dank gilt James Gasson, dessen Kooperation wesentlich die kreativen Momente dieser Arbeit gefördert hat. Ferner hätte ich die praktischen Ex- perimente nicht ohne die tatkräftige Unterstützung vieler Menschen am IKFT durchführen können. Stellvertretend möchte ich hier Birgit Rolli, Thomas Tietz, Matthias Pagel, Sonja Habicht und Armin Lautenbach nen- nen. Auch ohne die Zuarbeit vieler Studenten wäre diese Arbeit in den let- zten Jahren nicht bis zu diesem Punkt gereift. Herzlich danken möchte ich deshalb David Steinbach, Tatjana Sutter, Jonathan Jeras, Philipp Kempe, Saskia Lamour, Lucile Courtinat, Yichie Chien, Yali Liu, Matthew Schu- macher, Stephanie Hauschild und Alvaro Herreras. Sie alle haben maßge- blich zum Gelingen dieser Arbeit beigetragen. Schließlich möchte ich meiner kleinen Familie danken, die mich täglich getragen hat und immer die men- tale Stütze war, auch wenn an manchen Punkten Unmut, Demotivation oder Verzweiﬂung geherrscht haben. Meine Frau Raquel und mein ungeborenes Kind haben mir immer die Kraft gegeben, die ich brauchte, um diese Arbeit bis zum Ende zu führen. Abstract

The recovery of platform chemicals from biomass gains a growing govern- mental, economic and public interest. This work deals with the hydrothermal liquefaction of lignin. The focus laid on the discovery of the main reaction pathways within the depolymerisation of the lignin-macromolecules, the discovery of bottleneck reactions and the investigation of reaction mechanisms. In this context also the influence of the solvent as well as the influence of catalysis were studied. This was achieved by intensive literature study and experimental work in both batch and continuous reactor. Simulation of the thermodynamic equilibrium as well as modelling of the formal kinetics of the discovered reactions facilitated the evaluation of experimental results. The aim of this work was to provide tools for the optimization of process parameters and to indicate the direction for further studies regarding the conversion of lignin into phenolics. Thus, the economic feasibility of different process scenarios was studied in order to evaluate the attractiveness of a technical process.

Solvent studies involved the comparison of hydrothermal lignin depolymerisation and lignin solvolysis in ethanol. Lignin depolymerisation was found to be faster in water than in ethanol at moderate temperatures (633 K). Furthermore, the comparison of both solvents, water and ethanol, showed that water leads to a narrower spectrum of phenolic products and a tenfold increase of the catechol yield. This is advantageous in respect to the separation of the phenolic product from solvent and by-products. The simulation of the thermodynamic equilibrium with Aspen Plus V7.2 showed that the major part of any organic input material including 100 % of the solvent ethanol at 633 K would be converted into gaseous components while water showed to be resistant against thermal decomposition. Heterogeneous catalysts showed very positive eﬀects on the hydrodeoxygenation of primary or secondary phenolic products from hydrothermal lignin depolymerisation. The application of Raney-Ni e.g. selectively catalyses the conversion of catechol into phenol. Among the phenolic products recovered from the lignin degradation phenol is currently most utilized in the chemical industry. Kinetic modelling aided in the validation of the a priori model of reaction pathways and the discovery of bottleneck reactions within the lignin depolymerisation. It was proven that certain reactions can be predicted by the origin of the lignin, or the utilized solvent. Especially the inﬂuence of the chosen reactor was elucidated. Employing a Continuous Stirred Tank Reactor (CSTR) facilitates capping reactions, which leads to reduced yields of solid residue and gaseous products. At the same time it favours the formation of phenolic

i products. Elucidating reaction mechanisms contribute to the explanation of these phenomena. The production of phenol from lignin by solvolysis in ethanol or hydrothermal degradation is economically not feasible if the current state of the art is considered. Studies of diﬀerent scenarios showed that yields of phenol and other phenolics from lignin should be increased from at most 10 % to at least 20 % while the lignin/solvent ratio must increase and the mean residence time must be reduced in order to reduce especially the capital related costs.

Zusammenfassung

Die Gewinnung von Plattformchemikalien aus Biomasse bekommt wach- sende Aufmerksamkeit seitens Regierungen, Wirtschaft und Öffentlichkeit. Diese Arbeit beschäftigt sich mit der hydrothermalen Verflüssigung von Lignin. Der Fokus liegt auf der Entdeckung der Hauptreaktionspfade, welche die Depolymerisation der Ligninmoleküle umfasst, vor allem der kritischen Reaktionen sowie der Erforschung der Reaktionsmechanismen. In diesem Zusammenhang werden auch der Einfluss des Lösungsmittels und der Kata- lyse diskutiert. Dies sollte erreicht werden durch intensive Literaturstudien und Laborexperimente im Batchreaktor und im kontinuierlichen Rührkessel. Die Simulation des thermodynamischen Gleichgewichts sowie das Model- lieren der Formalkinetik der postulierten Reaktionspfade zeigten sich als probate Mittel für die Evaluierung der Versuchsergebnisse. Das Ziel der Arbeit war es, Werkzeuge für die Optimierung der Prozessparameter verfüg- bar zu machen und die Richtung für weiterführende Studien der Ligninver- flüssigung aufzuzeigen. Dazu gehörte auch eine Machbarkeitsstudie, welche das wirtschaftliche Potential eines technischen Prozesses zur Gewinnung von Phenolen aus Lignin evaluiert.

Zur Untersuchung des Lösungsmitteleinflusses wurde ein Vergleich zwischen hydrothermaler und solvolytischer Spaltung in Ethanol herangezogen. Die Lignin-Depolymerisation war bei moderaten Temperaturen (633 K) in Wasser schneller als in Ethanol. Ferner zeigte der Vergleich zwischen Wasser und Ethanol, ein engeres Produktspektrum bei hydrothermaler Spaltung von Lignin. Gleichzeitig stiegen die Catechol-Ausbeuten um das Zehnfache. Dies bietet vor allem Vorteile in Bezug auf eine Produkttrennung von Lö- sungsmittel und Nebenprodukten. Eine Simulation mit Aspen Plus V7.2 zeigte, dass der größte Teil der organischen Komponenten inklusive 100 % des Lösungsmittels Ethanol im thermodynamischen Gleichgewicht bei 633 K in Gas umgewandelt würden. Wasser hingegen ist sehr resistent gegen ii thermische Zersetzung. Heterogene Katalysatoren zeigten positive Effekte bezüglich der Hydrodeoxygenierung von primären oder sekundären phenolischen Produkten der hydrothermalen Lignin-Depolymerisation. Der Ein- satz von Raney-Ni katalysierte beispielsweise selektiv die Umwandlung von Catechol in Phenol. Unter den phenolischen Produkten, die bei der Lignin- zersetzung entstehen, ist Phenol derzeit die gängigste Plattformchemikalie in der chemischen Industrie. Kinetische Studien halfen, die a priori postulierten Modellreaktionen zu validieren und die kritischen "Flaschenhals"-Reaktionen ausfindig zu machen. Es konnte gezeigt werden, dass verschiedene Reak- tionen vorhergesagt werden können, wenn beispielsweise die Herkunft des Lignins oder das eingesetzte Lösungsmittel berücksichtigt werden. Beson- deren Einfluss hat auch der Einsatz verschiedener Reaktortypen. Ein kon- tinuierlicher Rührkessel erleichtert die Sättigung von Radikalen, was zu einer reduzierten Ausbeute von festen Rückständen sowie gasförmigen Produk- ten führt. Gleichzeitig wird die Entstehung von phenolischen Produkten gefördert. Wesentliche Erkenntnisse über die Reaktionsmechanismen trugen zur Erklärung dieser Phänomene bei. Die Phenolgewinnung aus Lignin durch hydrothermale oder solvolytische Spaltung in Ethanol ist unter Berücksich- tigung des aktuellen Stands der Technik wirtschaftlich nicht machbar. Die Studie verschiedener Szenarien zeigte, dass die Ausbeuten von phenolischen Produkten von etwa 10 % auf mehr als 20 % steigen sollten, wobei gleichzeitig das Lignin/Lösungsmittel-Verhältnis deutlich steigen und die Verweilzeit im Reaktor deutlich gesenkt werden muss um vor allem die Kapitalkosten für die Anlage zu senken.

iii iv Contents

Abstracti

Glossaries ix Glossary...... ix Acronyms...... ix Symbols...... xi

1 Introduction1 1.1 Motivation...... 1 1.2 Background and state of the art...... 4 1.2.1 Lignin...... 4 1.2.2 Chemical reaction mechanisms...... 7 1.2.3 Solvent...... 8 1.2.4 Kinetic studies...... 9 1.2.5 Catalysis...... 10

2 Materials, methods and analysis 13 2.1 Materials...... 13 2.2 Methods...... 14 2.2.1 Microbatch-autoclaves...... 14 2.2.2 Continuous stirred tank reactor...... 15 2.2.3 Delplot analysis...... 17 2.3 Analysis...... 17 2.3.1 Gas chromatography (GC)...... 17 2.3.2 Fourier transformed infra red spectroscopy (FT-IR). 19 2.3.3 Karl-Fischer-titration (KFT)...... 19 2.3.4 Photometry...... 20

v 3 Thermodynamic Studies 21 3.1 Introduction...... 21 3.2 Input parameters...... 22 3.3 Results...... 23 3.4 Discussion...... 26

4 Inﬂuence of solvent on thermal lignin degradation 29 4.1 Introduction...... 29 4.2 Experimental Results...... 29 4.3 Discussion...... 34

5 Inﬂuence of catalysts on thermal lignin degradation 37 5.1 Introduction...... 37 5.2 Screening of heterogeneous catalysts...... 38 5.3 Raney-Nickel...... 39 5.4 Discussion...... 42

6 Kinetic studies of lignin degradation 45 6.1 Lignin depolymerisation in ethanol in a batch-reactor.... 46 6.1.1 Main intermediates and products...... 46 6.1.2 Grouped products...... 50 6.1.3 Kinetic model development...... 53 6.1.4 Results from model ﬁt...... 57 6.1.5 Discussion...... 59 6.2 Lignin depolymerisation in ethanol in a continuous reactor.. 62 6.2.1 Experimental...... 62 6.2.2 Results...... 63 6.2.3 Discussion...... 73 6.3 Lignin depolymerisation in water in a batch-reactor...... 76 6.3.1 Model adaptations...... 78 6.3.2 Results...... 79 6.3.3 Discussion...... 83

7 Feasibility study of lignin liquefaction 93 7.1 Background Scenario...... 93 7.2 Process design...... 94 7.3 Main equipment...... 96 7.4 Cost estimation...... 100 7.4.1 Market prices...... 100 7.4.2 Manufacturing costs...... 102 7.5 Discussion...... 105 vi 8 Conclusion 111

Bibliography 117

List of Tables 129

List of Figures 133

A Materials 139

B Sample workup 145 B.1 Analysis overview...... 145 B.2 Extraction...... 147

C Modelling 151 C.1 Optimisation function...... 151 C.2 Solvolysis in ethanol...... 153 C.3 Solvolysis in water...... 156 C.4 Analysis of error...... 159 C.5 Sensitivity analysis...... 159 C.6 Flux analysis...... 160

D Extended studies of catalysts 163 D.1 Homogeneous catalysts...... 163 D.2 Catalytic hydrothermal decomposition of catechol...... 165 D.2.1 Results...... 165 D.2.2 Discussion...... 170

E Experimental results 173 E.1 Tables of experimental results...... 173 E.2 Energy and mass ﬂow...... 194

F Cost oﬀers 197

vii viii Glossar substances which are liquid at room temperature or soluble in aliphatic the solvent employed for the hydrocarbon molecule in which treatment. the carbon atoms are linked in open chains. methoxyphenols aromatic guaiacol derivations, include hydrocarbon molecule comprised methyl and ethyl para- of at least one benzene-ring. substitutions for guaiacol and syringol and propyl para- substitutions for syringol. bioenergy energy obtained through the con- O/C-ratio version of biomass into gaseous, molar quantity of oxygen divided liquid or solid fuels or by e.g. by the molar quantity of car- combustion of biomass. bon comprised by an organic sub- biofuel stance. fuel derived from organic matter, obtained directly from plants, or phenolic indirectly from agricultural, com- phenol-derived substance commercial, domestic, and/or indus- prised of a single benzene-ring. trial wastes, instead of from fossil phenols products. phenol derivations, include all methyl and ethyl substitutions for catechols phenol. catechol derivations, include pyrolysis methyl and ethyl para- most authors use the term if the substitutions for catechol. mixture of reactants consists of maximum 50 wt% organic sol- H/C-ratio vent. molar quantity of hydrogen divided by the molar quantity of solvolysis carbon comprised by an organic reaction of solutes, that are in substance. this case lignin and lignin derived components, and the solvent, that hydrothermal is ethanol. Water is potentially description of the treatment of inert. a substance at increased temperatures suﬃcient to break down solvothermal the chemical structure of the sub- description of the treatment of stance in the presence of an aque- a substance at increased temper- ous solvent. atures suﬃcient to break down the chemical structure of the substance in the presence of an or- liquefaction ganic solvent. conversion of solid biomass into

ix Acronyms ILUC indirect land use change

AA acetic acid K potassium Al aluminium KFT Karl-Fischer-titration

BDE bond dissociation enthalpy LHV lower heating value BET Brunauer, Emmett, Teller Lig lignin LLE liquid-liquid equilibrium C carbon MA micro batch autoclave CatOH catechol MeCatOH methylcatechol Co cobalt MeGua methylguaiacol COM total manufacturing costs MePhe methylphenol Cr chromium methoxyPhOH methoxyphenols CSTR Continuous Stirred Tank Reac- Mo molybdenum tor MS mass spectrometry CyclohexO cyclohexanone MtOE million tons of oil equivalent CyclohexOH cyclohexanol N nitrogen DME dimethyl-ether n.a. not available DTG derivative thermogravimetry n.d. not detected Nb niobium EoS equation of state NCW near critical water EtCatOH ethylcatechol Ni nickel EtGua ethylguaiacol EtOH ethanol O oxygen EtPhe ethylphenol OLC operating labor costs EU European Union Exp. experiment P phosphor Ext extraction P&ID piping and instrumentation diagram FA formic acid Pd palladium Fe iron Phe phenol FID ﬂame ionisation detector PhOH phenols FT-IR Fourier transformed infra red Pt platinum spectrometry Rh rhodium Ru ruthenium GC gas chromatography GLE gas-liquid equilibrium S sulphur GPC gel permeation chromatography SCW supercritical water Gua guaiacol Si silicon Syr syringol H hydrogen HC hydrocarbon Ta tantalum He helium TBM total bare module costs HHV higher heating value TCD thermal conductivity detector x TCI total capital investment H total dynamic head TGA thermogravimetric analysis H˙ i enthalpy stream of component i Ti titanium ∆H reaction enthalpy TMC total material costs R TUC total utility costs kj formal kinetic rate coeﬃcient of nonelemental reaction j

UV-Vis photometric spectroscopy LD depolymerised lignin, lump component summarising all lignin derived organic compounds soluble Symbols in the solvent, that is water or ethanol, and not measured by gas AEx heat transfer area chromatography (GC)-analysis

Aj constant describing the threshold ~k vector containing the formal kinetic value of kj approaching infinite rate coefficients of all reactions temperatures, apparent Arrhe- defined within the model nius factor m reaction mass Ci,aq,0 concentration of the component m˙ mass flow through the continuous re- i in an aqueous solution before ex- actor traction. mi mass of component i in the product Ci,org concentration of the component i mixture in the organic phase after extrac- mi,aq,0 mass of the component i diluted tion. in an aqueous solution before ex- cp average heat capacity. traction.

EA,j constant considering the temper- mi,org mass of the component i diluted ature dependence of kj , apparent in the organic phase after extrac- activation energy tion. mLig,0 initial mass of lignin fa factor considering the dilution of the organic phase after the ex- p reaction pressure

traction with ethyl acetate and pin initial pressure before entering the the internal standard pentacosane pump (C H ) 25 52 pout pressure of the effluent of the pump fb factor considering the solubility of q˙ heat transfer rate solvent and extractant R ideal gas constant fc factor considering the change of the ρ density solute concentration due to the ρ mean density of a mixture different input volumes of aqueous sample and extractant rj reaction R2 coefficient of determination fd,i factor considering the distribution ratio of the solute i in the solvent σ standard deviation and the extractant t experimental duration g gravitational acceleration τ residence time in a batch reactor

xi τ 0 mean residence time in the continu- w weighing factor ous reactor W work, energy T temperature Wo usable work Tc,W critical temperature of water x overall conversion ∆T difference of fluid temperature en- Ξ number of experiments tering and emerging the heat exchanger ∆yi sensitivity calc ∆T mean overall temperature dif- yi yield of the component i calcu- ference between the two fluid lated by employing the adequate streams exchanging heat model yi yield of the component i Tin temperature of fluid entering the heat exchanger yi,j yield of the component i taking part in the reaction rj ∆Tin temperature difference between meas hot fluid emerging the heat ex- yi,ξ yield of the component i mea- changer and cold fluid entering sured in the product of experi- the heat exchanger ment ξ

Tout temperature of fluid emerging the yi average of the yield of the compo- heat exchanger nent i measured in the product of identical experiments ∆Tout temperature difference between meas hot fluid entering the heat ex- yi average of the yield of the com- changer and cold fluid emerging ponent i measured in the product the heat exchanger of all experiments carried out at the same temperature Vaq volume of aqueous phase y0 yield of the component i after modi- V˙ volumetric flow rate, capacity i fication of kj Vorg volume of organic phase yi,0 initial yield, that is the mass of the VR reaction volume component i divided by mLig,0

xii Chapter 1

Introduction

1.1 Motivation

Earth’s resources of crude oil are limited [OR11]. An important challenge for scientists and engineers is to develop technologies that are largely independent from fossil crude oils. Biomass is supposed to have a high potential to replace crude oil as a basic input material for the production of many organic chemicals and fuels. Political efforts to increase the market share of biomass derived chemicals and fuels in USA and Europe increase. »The European Union (EU) as a whole has set a mandatory target of 20 % for renewable energy’s share of energy consumption by 2020 and a mandatory minimum target of 10 % for biofuels for all member states« [Zak+0]. A study carried out by the European Environment Agency reports that a large biomass production capacity is available in Europe, which could produce 190 million tons of oil equivalent (MtOE) of biomass by 2010 with possible increases up to 300 MtOE by 2030 [EEA06]. However, the study »disregards the effect of competition between bioenergy and food production for domestic food supply.« This »would become more important with the assumed rise of the combined carbon permit and energy prices.« In fact, the study considered that »increased demands on agricultural sector output causes intensification of farm management across the agricultural land area, incentives to trans- form extensively used grassland, olive groves or dehesas [...], into arable land for growing bioenergy crops« and »an inappropriate bioenergy crop mix, which does not take account of the specific environmental pressures of different crops in the context of the main environmental problems in a particular region« [EEA06]. However, it could still not be proven that these

1 1.1. Motivation worrisome effects of increasing biomass production can be limited by the current policy and the agricultural industry. On the contrary, the worldwide agricultural production combined with »intensive export-oriented agricul- ture has increased under open market operations but has been accompa- nied by both benefits and adverse consequences depending on circumstances such as exportation of soil nutrients and water, unsustainable soil or water management or exploitative labor conditions in some cases« [Bei+08]. In addition, »people have benefited unevenly from these yield increases across regions« [Bei+08]. Furthermore, the increase in bioenergy in Europe is sus- pected to foment e.g. the grubbing of tropical rainforests. These effects of indirect land use change (ILUC) are still discussed, however, »simulations forEU biofuels consumption above 5.6 % of road transport fuels show that ILUC emissions can rapidly increase and erode the environmental sustain- ability of biofuels« [ADL10]. The recent effects of common agricultural and renewable energy policy in Germany show that e.g. the area utilised for the cultivation of maize has risen of approximately 50 % to 2.3 million hectare in the decade from 2000 to 2010 due to high market prices and subsidies [SR11]. 23 % of this area was dedicated to the generation of bioenergy in 2011 [FNR11; Ehr12]. Maize, however is ranked to highly contribute to soil erosion nutrient, leaching to ground and surface water, pesticide pollution and reduction of farmland biodiversity [EEA06]. Should our society rely on the increasing output of the agricultural industry to satisfy its energy demand? The fields of discussion are large and will not be entirely covered here. In order to establish neither the generation of energy nor the production of chemicals from biomass at the expense of biodiversity or social and environmental pressures, within the present work the use of biomass exclusively from waste is proposed. The potential of waste biomass, though hopefully decreasing in the future, is 100 MtOE per year [EEA06]. Lignin, a major component of lignocellulosic biomass, 15-30 % by weight, 40 % by energy [Zak+0], is one of the most abundant renewable organic materials in the world. Large quantities of lignin are available as waste material produced by the pulp and paper industry (black liquor). Worldwide more than 50 million tons of dry lignin are produced annually, of which more than 98 % are burned to recover its energetic value [Gla09; Pul09]. Less than 2 % are used for synthesis of fuel or chemicals. Upcoming technologies are the gasification of black liquor and adjacent synthesis of dimethyl-ether (DME) or Fischer-Tropsch-synthesis [CKL09]. Chemrec is already running a gasification plant in Piteå (Sweden) [Sch12]. From a chemical perspective, lignin is a readily available source of aromatics and phenolics that are present within the structure [KB08a], which at present is only marginally being

2 Chapter 1. Introduction used for production of, e.g. vanillin [KT89], environmental-friendly glue [Sch07] and in lead batteries [Kog09]. Aromatic and phenolic compounds are, however, of great interest for use as platform chemicals for, e.g. the production of synthetic resins or pharmaceuticals, and could possibly be used as an unresolved mixture for fuel blending purposes, based on the anti oxidative properties of hindered phenolics in turbine fuels [AST87]. Among them phenol is the most demanded chemical substance. In 2004 8.5 Mt/a were produced worldwide [Arp07]. Large research effort was inverted in the study of lignin depolymerisation and the production of fuel or chemicals from lignin. In the middle of the last century Lautsch and Piazolo carried out catalysed hydrogenation of lignin in a basic solution and alcohol [LP43]. The investigation continued. A compre- hensive summary was available in 1981 written by Meier and Schweers [MS1]. Towards the end of the last century research efforts concentrated on pyrolysis of lignin without solvent (dry pyrolysis) or in organic solvents [FMG87; MB97; APR01]. Most studies which try to elucidate reaction pathways of hydrothermal lignin degradation were conducted in the last decade [WSG11; YM12]. However, no technical process for the production of phenolics from lignin has been released. This might be due to relatively small yields of phenolics from lignin, usually around 10 to 20 wt% maximum [APR01]. The theoretically possible yield is, depending on the type of lignin, between 30 and 50 %. In order to increase the yield of phenolics, catalytic hydrogenation of the liquid product from lignin pyrolysis was applied [Gut+09; Kar+06]. However, most research efforts focus on model substances and disregard real product mixtures which comprise a large variety of single compounds. In order to understand how yields of phenolics from thermal-chemical lignin degradation can be increased, it is necessary to elucidate important reactions and their participants which occur within the lignin depolymerisation. The intention is to reveal bottleneck reactions and main reaction products. It is thus also necessary to understand reaction mechanisms and how they are influenced by several parameters like temperature, solvent, catalyst, etc. The present work approaches the answers to these questions by focussed experimental work. Simulation of the thermodynamic equilibrium and modelling of the reaction kinetics are strong numerical instruments for the interpretation of the experimental results. The aim of lignin depolymerisation is to maximise the yield of valuable phenolic products by optimising reaction parameters. The influence of different solvents and catalysts on the phenolic yield from thermal-chemical lignin degradation are discussed in Chapter4 and5. The simulation of the thermodynamic equilibrium is especially interesting for the evaluation of the influence of the residence time in the reactor and temperature on the

3 1.2. Background and state of the art product composition (see Chapter3). For the recovery of valuable phenolic products from lignin, it is crucial to understand the main reaction pathways of lignin degradation and reaction kinetics. Up to now, some research has been undertaken in respect to the kinetics of pyrolytic, solvothermal and hydrothermal lignin depolymerisation [KV08; PK08; Var+97; ZHR11; YM12]. The present work includes the development of a formal kinetic model considering the formation and degradation of the main phenolic products as well as the solid and gaseous byproducts (See Chapter6). The development of a model of the crucial reaction pathways and the kinetic studies allow a suggestion of these optimised parameters and clear statements about the basic design of an industrial process of lignin depolymerisation. A feasibility study of a technical process for thermal-chemical degradation of lignin in both, water and ethanol, was considered to be essential for the evaluation of future research work concerning the production of phenol from lignin (see Chapter7).

1.2 Background and state of the art

The following sections provide an overview over the state of knowledge and technology concerning the molecular structure of lignin and the methods of separation from the cellulosic parts of biomass (see Section 1.2.1), reactions occurring within thermal-chemical lignin degradation and their mechanisms (see Section 1.2.2), specific solvents used within this work and their effects on lignin degradation (see Section 1.2.3), kinetic studies (see Section 1.2.4) and the influence of homogeneous and heterogeneous catalysts (see Section 1.2.5).

1.2.1 Lignin Degradation of lignin is a complex process. Not only a vast spectrum of compounds can be produced from lignin, but lignin itself is a largely varietal polymer. The origin of the biomass largely contributes to the structural-chemical variations encountered in different types of lignin. Different types of lignocellulosic biomass consist of different amounts of lignin, with lignin abundance generally decreasing in the order softwood > hardwood > grass/annual plant. In plant cell walls, lignin fills the spaces between cellulose and hemicellulose, and it acts like a resin forming the matrix where fibrous cellulose is embed- ded [Zak+0]. Lignin is a three dimensional amorphous polymer consisting of methoxylated phenyl-propane building blocks (see Figure 1.1). Varying amounts of the three characteristic phenyl-propane building blocks, shown in Figure 1.2, allow its classification into different types of lignin and subse-

4 Chapter 1. Introduction quently into the three broad categories of plant lignin: softwood, hardwood, and grass or annual plant lignin [Pea67, p 339].

Figure 1.1: Example of lignin structure depicting some of the dominant linkages between the aromatic monomers [WR12].

Table 1.1: BDE of the most predominant linkages (see Figure 1.1.) Bond α-O-4 β-O-4 4-O-5 β-1 α-1 5-5 BDE (kJ/mol) 215 290 330 280 360 490

Softwoods are commonly composed of guaiacyl lignin, which is principally made up of coniferyl moieties. Guaiacyl-syringyl lignin contains a combination of sinapyl and coniferyl moieties and is prevalent in hardwoods. Annual plant or grass lignin is mainly composed of para-coumaryl units [FN68]. The characteristic variation of these diﬀerent units can also be observed in the depolymerised products [Dor+99]. The building blocks (aromatic monomers)

5 1.2. Background and state of the art

Figure 1.2: The three characteristic aromatic substances which constitute the characteristic monomeric units of lignin: A: para-coumaryl alcohol, B: sinapyl alcohol and C: coniferyl alcohol.

are randomly connected through C-C bonds or ether bonds (C-O bonds). The most dominant bonds (80 %) within the lignin structure are ether bonds [RS09]. The distribution of the linkages in the biopolymer structure depends on the type of wood and the part of the plant. The most predominant ether linkages are the β-O-4 and the α-O-4 [WR12]. These and other linkages are depicted in Figure 1.1. The BDE, listed in Table 1.1, reveals that the ether bonds in general are weaker than the C-C bonds. Hence, the cleavage of the ether linkages is the most favourable reaction initiating the depolymerisation of the lignin polymer.

Furthermore, the structure of the lignin separated from the rest of the plant structure (cellulose and hemicellulose) depends on the employed separation method. The separation method can consist in physical treatment, solvent fractionation, chemical or biological treatment [Cos+9]. Alkaline treatment, a chemical treatment, is most commonly employed by the pulp and paper industry. It is a highly efficient separation method, however, in- volves the incorporation of sulphur which can be problematic for a further chemical treatment especially if heterogeneous catalysts are involved. Var- ious novel developments within the most prominent separation techniques, i.e. enzymatic and acid hydrolysis, organosolv pulping and steam explosion have largely contributed to establishing a pathway to a potentially economically valid exploitation of the sugar polymers from lignocellulosic biomass. As a side-effect, they further shape some of the characteristics of the lignin by reacting with functional end-groups in the lignin structure, thus influencing the solubility properties of the isolated polymer.

6 Chapter 1. Introduction

1.2.2 Chemical reaction mechanisms By the cleavage of ether bonds, the main linkages between the aromatic rings of the lignin macromolecule, aromatic monomers and oligomers are formed. It is however a challenge to gain a high-value product of homogeneous composition from such a chemically complicated and inhomogeneous component as it is lignin. Hence, it is helpful to understand at least the main chemical mechanisms that are responsible for the most important reactions occurring within lignin depolymerisation. Formation and degradation of aromatic monomers occur by hydrolysis [TRR11] or radical reaction [DM99]. The most relevant initiation reactions involve the cleavage of the weaker C-O bonds as explained above. If ho- molytic cleavage and thus radical reaction is assumed, the BDEs are a key value. Faravelli et al. [Far+10] discussed diﬀerent radical reaction pathways which lead to initial cleavage of ether bonds and the BDEs of some of the bonds which are present in the lignin structure. Hydrolytic reactions involve the heterolytic cleavage of ether bonds, especially β-O-4. Possible hydrolytic reactions occurring within the degradation of lignin were summarised by Lundquist and Lundgren [LL1; Lun6]

Figure 1.3: Monoaromatic products (phenolics) from lignin depolymerisation: A: phenol, B: catechol, C: guaiacol and D: syringol.

The cleavage of the ether and aryl bonds of the lignin structure yields monoaromatic (phenolic) substances. The most prominent are phenol, catechol, guaiacol and syringol (depicted in Figure 1.3) and alkylated derivatives thereof. The phenolics further decompose by thermal cleavage of the hydroxyl- and methoxyl-group [Run+12; Nim+11a; Nim+11b]. At the temperatures usually applied for liquefaction of lignin (500 - 750 K) the aromatic ring is stable. It decomposes towards higher temperatures [HS2]. The C-O bond of the methoxy-group of guaiacols and syringols is most likely to decompose due to rather low BDE of about 240-250 kJ/mol [Vuo86]. Diﬀerent pathways of the decomposition of methoxyphenols are comprehensively summarised by Zakzeski et al. [Zak+0].

7 1.2. Background and state of the art

The recombination of fragments from both radical [McM+04] and hydrolytic [Sai+03] cleavage of lignin ether bonds leads to the build-up of a polymer. This polymer is usually referred to as char. Dorrestijn et al. claimed that char formation can be caused by hydrogen shuttling from the lignin matrix before the actual process temperature is reached, which implies that the structure of lignin under liquefaction conditions is irreversibly changed [Dor+00]. McMillen et al. postulated a repolymerisation of (di-) hy- droxybenzene which occurs during liquefaction of coal [McM+04]. Roberts et al. observed polymerisation of monomeric aromates during hydrothermal treatment of lignin and suggest boric acid as an excellent radical scavenging agent [Rob+1]. Some aromatic monomers, such as phenol or cresol, were also found to inhibit the formation of char and reduce the yield of solid residues by scavenging aromatic radicals and thus stopping the repolymerisation reaction [McM+04; Sai+03; Tak+12]. Yokoyama et al. [Yok+98] found that char formation from lignin in aqueous media was promoted by low temperatures, high pressures and long reaction times.

1.2.3 Solvent Pyrolysis, solvolysis and hydrothermal treatment are the most promising advances to lignin depolymerisation for the purpose of liquefaction and recovery of phenolics [Now+10; KGB09; Moh06]. Lignin depolymerisation and valorisation by hydrodeoxygenation has been widely explored within different pyrolysis and solvolysis approaches. Solvothermal and hydrothermal degradation provide the advantages of mild conditions and a single phase environment due to the miscibility of the organic products in the (supercritical) solvent. M. Kleinert and T. Barth have done intensive research on lignin solvolysis in ethanol using formic acid as hydrogen donor [KB08a; KB08b; Kle+11; KGB09]. Ethanol has a very high solvency for biomass and a low critical temperature which makes it attractive for lignin depolymerisation [Zha+10b]. Further advantages of solvolysis in ethanol over fast pyrolysis are a less oxygenated oil fraction [KB08b] and almost no solid residue (< 5 wt%) [Gel+08; Moh06]. In-situ hydrogen donation is realised by the thermal degradation of formic acid. A very active atomic hydrogen species is produced [Gel+08], which results in a successful deoxygenation and hydrogenation of lignin structures. Various largely demethoxylated phenolic structures comprising a high H/C-ratio and low O/C-ratio[KB08b] were obtained without necessitating any further added catalyst for activation or transfer. The selection of a solvent for thermal-chemical lignin depolymerisation is one of the key challenges since it has a great impact on the deoxygena-

8 Chapter 1. Introduction tion of aromatic intermediates as well as on the formation of char and gas. Liu and Zhang [LZ08] compared water with two organic solvents, that are ethanol and acetone. Increased char yields have to be expected when using water probably due to the higher solubility of oily and tarry products in the organic solvents. However, water shows a narrower product spectrum in respect to phenolic products and a high selectivity in respect to phenol and catechol [WSG08]. This is an advantage especially for the separation of the products, which tends to be less energy and cost intensive at increased concentrations of the target product. The yield of catechol and phenol can be even increased by applying an acidic or a basic environment [TRR11]. Wahyudiono, Sasaki, and Goto [WSG08] pointed out the inﬂuence of the water density on the product composition. High water density led to higher phenol and cresol yields between 623 K and 673 K. Further advantages of water are its abundance at rather low prices and its chemical stability at the required temperatures and pressures. In fact in respect to the treatment of black liquor no drying prior to the liquefaction has to be applied, thus energy can be saved.

1.2.4 Kinetic studies The understanding of the degradation processes and formation of products is an essential step in increasing yields and achieving product requirements. This requires both the elucidation of pathways but also of their kinetics. Ki- netic models describing the decomposition of lignin are largely concentrated on global bulk models describing the yields of gas, char and liquids. First attempts describing the degradation were done using a single first order reaction [Nun+85], which were followed by more complex models, involving also multiple competing reactions [Var+97]. A semi-detailed model based on a set of molecular and radical species derived from the main building blocks involved in about 500 elementary or lumped reactions considering the cleavage of different bond-types and repolymerisation reactions was developed by Faravelli et al. [Far+10]. During the last decade the thermal degradation of lignin was studied by various authors applying thermogravimetric analysis [Cho+12; JNB10]. In addition, the kinetics of hydrothermal lignin degradation came in the focus of research. Zhang, Huang, and Ramaswamy calculated formal kinetic parameters of the hydrothermal degradation and the build-up of char in a batch reactor [ZHR11]. Takami et al. applied a Monte-Carlo-Simulation in order to determine the formal rate coefficient of lignin degradation [Tak+12]. The first kinetic model considering the formation and decomposition of phenolics within the lignin degradation in a continuous reactor was reported recently by Yong and Matsumura [YM12].

9 1.2. Background and state of the art

Experimental studies of the decomposition and charting of network pathways under pyrolysis conditions were reported by Klein and Virk [KV80; JK85; KV08]. They considered the presence of hydrogen and exploited the available intermediate model substances as starting materials, such as syringol, guaiacol and catechol, to try and master the cascade of demethoxylation, deoxygenation, hydrogenation, alkylation, demethylation and methanolisa- tion reactions. The kinetics of the decomposition of both lignin and biomass degradation and different modelling approaches, the latter mainly focusing on global bulk models, have been comprehensively summarised by Miller and Bellan, Prakash and Karunanithi and Brebu and Vasile [MB97; PK08; BV10]. The intention of this work is to establish a model, that considers the main reaction participants and pathways and is able to quantify the amount of phenolic products in dependence of the residence time in the reactor. In addition, bulk products such as char and gas should be quantified respecting the mass balance at every time. A validation of the developed model using literature and employing different reactor systems was conducted. Finally, an adaptation of the model for different solvent systems was discussed.

1.2.5 Catalysis Both homogeneous and heterogeneous catalysts can influence the reactions occurring during thermal-chemical lignin degradation. While heterogeneous catalysts are more likely to interact with diluted compounds with a relatively low molecular size, such as phenolics, homogeneous catalysts are capable to influence also the cleavage of the ether and alkyl linkages of the lignin molecule [Beh+06]. Toor, Rosendahl, and Rudolf [TRR11] have summarised the effects of homogeneous and heterogeneous catalysts on the hydrothermal liquefaction of biomass in general. According to them alkali salts increase the liquid yield and inhibit the dehydration of the biomass monomers and thus suppress the char formation. Karagöz et al. found that K2CO3 also effects the formation of monoaromatic substances and increases the amount of di-hydroxybenzenes (catechols) from hydrothermal liquefaction of sawdust from pine at 553 K [Kar+06]. In addition, alkali salts, especially potassium (K) salts, have a high ability to catalyse the water-gas-shift reaction and thus increase the amount of CO2 and H2 at the expense of CO via formate as intermediate. The presence of reactive hydrogen potentially promotes the hydrogenation of organic intermediates [Kru+00; SKR04]. During hydrothermal liquefaction some gasification is crucial since the oxygen is removed under the formation of CO2. However, extensive gasification will reduce the yield of liquid products [TRR11]. Sergeev and Hartwig claim to have found

10 Chapter 1. Introduction a homogeneous nickel carbene catalyst which selectively cleaves aromatic C- O bonds in various aromatic ethers without reduction of aromatic rings or cleavage of aliphatic C-O bonds [SH11]. Heterogeneous catalyst were applied mostly for the gasification of biomass and rarely for direct liquefaction. However, much effort has been spent on the investigation of valorisation of bio-oils by hydrodeoxygenation. The bio- oils are obtained in a first step liquefaction of biomass and in a second step hydrodeoxygenated with the help of a heterogeneous catalyst in a hydrogen donating environment [Wil+09]. Dorrestijn et al. suggest the catalytic hydrogenation in an appropriate hydrogen donating solvent to be the most promising method to produce phenols from lignin [Dor+99]. Catalysts applied for hydrodeoxygenation are heterogeneous catalysts. Wild et al. used ruthenium (Ru) on carbon (C) for valorisation of lignin derived pyrolysis oil [Wil+09]. Other studies show that catalysts based on rhodium (Rh), palladium (Pd) and platinum (Pt) have a high potential to promote hydrodeoxygenation of methoxyphenols in organic solvents [Gut+09] as well as in a gas phase reaction [Zha+11]. The reaction network of catalytic hydrodeoxygenation and the influence of an acid, e.g. the support of a metal catalyst, on the oxygen removal were developed by Nimmanwudipong et al. [Nim+11a; Nim+11b; Run+12]. In addition, heterogeneous catalysts comprising nickel (Ni) and molybdenum (Mo) were studied and their effect on the hydrodeoxygenation of bio-oil derived components were proven [LWC11]. Noteworthy in this context is the statement of Rinaldi and Wang saying that the solvent holds the key to control the selectivity of a heterogeneous catalyst in the conversion of lignin into phenols [RW12; WR12]. They reported that Raney-Ni in combination with 2-propanol as solvent and H-donor ef- ficiently catalyses the hydrodeoxygenation of model substrates and bio-oil [WR12]. The influence of heterogeneous catalysts on lignin-derived phenolics and the hydrodeoxygenation of pyrolysis oil in an aqueous phase was studied by Zhao and Lercher [ZL12; ZL2; Zha+1] who suggest the conversion of phenolic components present in bio-oil in an aqueous solvent with Ni or Pd catalysts in combination with an acid, e.g. Nafion/SiO2 [Zha+10a]. The present work focuses on heterogeneous catalysis of the valorisation of phenolics in an aqueous solvent especially with Raney-Ni which showed promising effects on the conversion of catechols into phenols [For+12].

11 1.2. Background and state of the art

12 Chapter 2

Materials, methods and analysis

2.1 Materials

For the experimental work two diﬀerent types of lignin were used which are distinguished by the origin of the wood and the method of separation of cellulose and lignin. The ﬁrst type of lignin was received from ALM Indiva Pct. Ltd., India, and originates from wheat straw. The separation from the cellulosic parts of the wheat straw was realised in a basic aqueous medium, followed by a precipitation of the cellulose employing a weak acid. The second type of lignin was received from Sekab, Sweden, and originates from spruce which is a softwood. It was treated by enzymatic hydrolysis in order to separate lignin and cellulose. Both types of lignin were dried for at least 16 h at 378 K and ground to a maximum diameter of 0.2 mm. The residual moisture of both types of lignin after drying and storage was 4.5 wt%. The amount of residual cellulose and hemicellulose obtained by analysis and data provided by the producer are listed in Table 2.1 together with the results from CHNS analysis (see Section 2.3.4) yielding the percentage ofC, hydrogen (H), nitrogen (N), sulphur (S) and oxygen (O). Chemicals such as any heterogeneous or homogeneous catalysts, phenolic compounds for analysis and experiments, solvents and gases are listed in AppendixA.

13 2.2. Methods

Table 2.1: Characterisation of both lignin types from wheat straw and spruce. Type/ Lignin Cellu- Mois-CHONS Ash Origin losea ture wt% wt% wt% wt% wt% wt% wt% wt% wt% Wheat >90 <3 4.5 63.8 6.0 27.0 1.5 0.3 1.4 straw Spruce 40 47 4.5 60.5 6.7 30.1 0.6 0.1 2.0 a Including hemicellulose.

2.2 Methods

2.2.1 Microbatch-autoclaves

The most used type of reactors for the experimental work was an in-house custom built micro batch autoclave (MA) of 5 mL or 10 ml inner volume made of stainless steel 1.4571 (see Table A.3). It seals gas-tight up to temperatures of 723 K and pressures of 30 MPa. A simple version of theMA(MA1) permits the loading of the reactor with solids and liquids, however, ventilation and loading with gas is not possible. In order to enable the loading of theMA with gas, the design was modiﬁed and gas tubes and valves were attached (gas-input version,MA2). Mechanical drawings of both versions of theMA are to be found in AppendixA. The MAs are easy to handle, can resist high pressures and temperatures and are easy to repair and to replace.

Simple version (MA1)

TheMA1 were ﬁlled with solvent, lignin and if appropriate a certain amount of catalyst. The remaining volume was ﬂushed with nitrogen gas and thereafter the reactor was screwed tight (metal on metal sealing) in a custom built containment that seals up to pressure of 10 MPa. TheMA1 was heated in an oven (HP-5890) at maximum heating rate (approximately 40 K/min). The experiment was started when the desired temperature in the oven was reached. After the determined residence time τ at the temperature T the autoclaves were cooled either by active air cooling of the oven or if more rapid cooling was required in iced water. After cooling the micro-autoclave were again placed in the custom-build containment for ventilation.

14 Chapter 2. Materials, methods and analysis

Gas-input version (MA2) The further developedMA2 has one inlet-tube on the side of the bottom part of the autoclave and one outlet-tube connected to the top-part of the MA2 (see Figure A.2). Both tubes can be opened and closed with valves that seal gas-tight up to a pressure of 100 MPa gauge pressure. For safety reasons the reactor had to be ventilated with nitrogen (N2) at first. After ventilating with N2 for a time duration of 30 s, the nitrogen-supply was cut and the inlet-tube was connected to a hydrogen bottle. The autoclave was ventilated with hydrogen (H2), again for 30 s. Afterwards, the outlet-valve was closed and the appropriate pressure of H2 in the autoclave was adjusted. After pressing H2, or any other desired gas, into theMA2, the hand-valve was closed and theMA2 was placed in an oven (HP-5890) where it was heated to the desired temperature at maximum heating rate (approximately 40 K/min). After the determined residence time in the oven theMA2 was cooled to room temperature in iced water. For ventilation, the outlet tube of theMA2 was connected to the inlet of a high pressure gas sampling tube that had been ventilated with nitrogen before. The outlet of this gas sampling tube was connected to a gas-meter. Now the hand valve was opened so that the gas inside the reactor would fill the gas sampling tube and afterwards flow into the gas-meter to measure the amount of gas coming out of the MA2.

Sample workup The volume of gas exiting from the autoclave was measured with a gas flow-meter. A sample of the gas was analysed on a gas chromatograph equipped with a flame ionisation detector (FID) and a thermal conductivity detector (TCD) in order to determine the composition. After ventilation, the reactors were opened and the liquids and solids were removed. The solids were separated from the liquids via filtration and the weight of both phases was determined gravimetrically. For detailed sample work-up of aqueous and organic liquid samples see AppendixB. The analysis of experimental errors is based on replicate experiments and is further explained in Appendix C.4.

2.2.2 Continuous stirred tank reactor A CSTR was employed for thermal degradation of lignin in order to take advantage from the characteristic properties of this type of reactor, that are the back mixing and a high heating rate [Lev99, p.94ﬀ]. Back mixing enables the interaction of products with unreacted feed. The CSTR has an inner volume of 200 mL, is made of the Ni-based alloy Inconel625 (see

15 2.2. Methods

Table A.3) and is heated by incorporated heating cartridges purchased from Watlow (type Firerod). It is designed to bear 923 K and 100 MPa pressure. The stirrer is made of 1.4571 stainless steel. The transmission of the torque between the motor and the shaft of the stirrer is realised by means of a magnet clutch. The magnet coupling has to be maintained below the Curie- temperature (1041 K) to guarantee the magnetic properties of the material and the movement of the stirrer throughout the experiment. A piping and instrumentation diagram (P&ID) flow scheme of the laboratory plant is shown in Figure A.3. The feeding of the suspension composed of lignin, solvent and co-solvent into the reactor is realised via two screw presses, each with a maximum inner volume of 60 mL. The screw presses were purchased from Swagelok and made of 1.4571 stainless steel. The suspension of ethanol, formic acid and lignin is sucked into the screw press cylinder through a tube on the front end of the cylinder and pressed out through a tube at the lower part of the cylinder of the screw press. In order to avoid plugging by means of sedimentation in the feeding tubes, long and curved tubes in the construction were avoided (length of feeding tubes less than 250 mm, inner diameter: 5.2 mm). The pneumatic valves and the motor of both screw presses are controlled manually and independently from each other which enables a continuous feeding and facilitates a rapid filling of the screw press cylinder while at the same time low flow rates can be realised. Before starting an experiment, the reactor was pressurised and heated to the desired temperature. Throughout the heating period, the reactor was rinsed with ethanol by a third pump. Upon commencing the experiment, the feed suspension was filled into the reservoir. For mixing purpose the feed was continuously pumped in a circle from the bottom of the feed reservoir to the upper part of the reservoir. The screw press was filled by sucking the feed suspension in from the circulation tubes. After leaving the reactor, the fluid passed a cooler and a depressuris- ing system, consisting of two valves, a pressure measurement system and a proportional and integral controller (see Figure A.3). The product stream then entered a flash separator. In the separator gaseous products were separated from solid and liquid products. The volume of the gas produced was measured using a rotor gas flow meter (Ritter TG1). The latter can optionally be conducted through a gas sample tube. Gas samples can thus be taken at different intervals during the experiment. When the gas flow showed less than 5 % deviation for a minimum of three measurements within 60 min, the stationary state was claimed to be reached. Several liquid samples (including solid particles) were taken throughout the whole experiment by emptying the flash separator via a valve at the bottom of the container. Further work-up procedures were conducted analogue to the work-up described for the batch

16 Chapter 2. Materials, methods and analysis reactors (see AppendixB).

2.2.3 Delplot analysis The Delplot technique bases itself on the analysis of selectivity-conversion plots and can be used for both primary and higher rank products [BKB90]. The basic primary Delplot allows the separation of primary from non-primary reactions for any given reaction order. For a primary Delplot, the yield of a singular ith component (yi) divided by the overall conversion (x) is plot- 2 3 ted over x. Plotting yi/x over x gives a second rank Delplot, yi/x over x third rank and so on. The intercept on the y-axis stipulates the rank of the product. If the intercept in a primary rank Delplot of a component is ﬁnite, the product is a primary one. If the intercept on the y-axis is zero, then the component is not formed by a primary reaction. The interpretation of intercepts is analogous for higher ranks.

2.3 Analysis

In this section the employed analysis methods are introduced and described. Basic data about the utilised equipment and adjustments are given. Uncer- tainties of the analytically obtained results and error tolerances were determined by replicate experiments. If error bars are given in any diagram, they are based of the standard deviation σ (see Appendix C.4). 18

2.3.1 Gas chromatography (GC) GC was employed to quantify the gaseous products and the phenolic products in the liquid product phase. After injection the sample is vaporised in the injection port and carried by a mobile gas phase through a column. Due to the different interaction forces with the surface of the column each constituent of the sample has a characteristic retention time in the column. Thus the separation of the constituents can be achieved. A quantification of the substances emerging from theGC-column was conducted with either a flame ionisation detector FID or a thermal conductivity detector TCD. In the FID the eluent emerging from theGC-column is mixed with hydrogen and air a burned and ionised. The ions are detected by the collector elec- trodes. The FID facilitates the quantification of substances comprisingC. The TCD senses changes in the thermal conductivity of the column effluent and compares it to a reference flow of the carrier gas helium (He).

17 2.3. Analysis

GC analysis of gaseous samples

Gas phaseGC analysis was performed on an Agilent 7890A with a 2 m Molsieve 5A column in series with a 2 m Porapak Q column equipped with a FID front and TCD back detector. The system was controlled by an Agilent laboratory data system. Injections were carried out by manually injecting 100 µL of the ventilated gas from the reactor. Temperature program: Initial temperature 323 K for 22 min, then heating at 20 K/min to 423 K, kept for 15 min. Further heating at 50 K/min to 503 K, kept for 10 min. The injection port was at 523 K, the FID at 503 K and the pressure was kept constantly at 255 kPa. The equipment was calibrated by introducing varying amounts of a standard gas mixture. The system was checked on a weekly basis by introducing the standard gas mixture and recalibrated if necessary. The maximum tolerance of the measured value was 10 vol% in respect to the concentration of each gaseous compound in the standard mixture.

GC-FID analysis of liquid samples

QuantitativeGC-FID analysis of liquid products was performed on a HP 5890-IIGC equipped with HP auto sampler 5890, a 30 m Rtx-1MS dimethyl- polysiloxan column and FID. The system was controlled by an HP-Chem laboratory data system. Temperature program: Initial temperature 313 K for 6 min, then 5 K/min heating to 453 K, then with a rate of 30 K/min continued heating to 533 K, kept for 5 min. Further heating increase at 30 K/min to a final 573 K and kept for 12 min. The injection port had a temperature of 548 K and the FID was at 603 K. Only organic samples could be analysed on this way. For analysis of an aqueous product an extraction of the phenolic products from the aqueous phase had to be conducted to prepare the sample (see Appendix B.2). The organic liquid samples were diluted in a 1:2 ratio with a prepared standard solution of pentacosane in ethyl acetate (1002 mg/L) as an external standard. injected by the auto-sampler system. Mono-aromatic components were individually calibrated for quantification by running dilution series. The equipment was calibrated by introducing samples with four different determined dilutions of the qualified substances. The system was checked on a weekly basis by introducing a standard mixture with known concentrations of all quantified substances and recalibrated if necessary. The maximum tolerance of the measured value was 10 wt% in respect to the concentration of each substance.

18 Chapter 2. Materials, methods and analysis

GC-MS analysis of liquid samples

AGC equipped with a mass spectrometry (MS) was used for the first identification and the determination of the retention time of each substance. In aMS the eluent of theGC-column is ionised by impacting them with an electron beam. The ions are separated according to their mass-to-charge ratio in an analyser by electromagnetic fields. Hence, theMS elucidates the chemical structure of the molecule and aids in characterising the substance. QualitativeGC-MS analysis of liquid products was carried out on a Trace UltraGC coupled with a DSQ II quadrupoleMS detector from Thermo Sci- entific. The samples were diluted in a 1:100 ratio in dichloromethane and were analysed using split injection at 523 K (injector temperature) on a 25 m Ultra 2 silica column ((5 % phenyl)-methylpolysiloxane) from Agilent Tech- nologies. A constant gas flow rate of 1 mL/min and a split ratio of 100 with the following temperature program were applied: Initial temperature 313 K, increased at 6 K/min to 473 K, further increased with 8 K/min to a final 573 K, kept for 5 min. TheMS detector was operated in positive ionisation mode at 70 eV with an ion source temperature of 473 K.

2.3.2 Fourier transformed infra red spectroscopy (FT- IR)

Fourier transformed infra red spectrometry (FT-IR) was employed for the analysis of solid samples. It aids in analysing the molecular structure, especially the dominant functional groups of the solid. FT-IR-analysis of the feed material and collected solid residues was performed by preparing and pressing conventional KBr pellets and analysing these on an FT-IR Varian 660-IR in linear transmittance mode. 30 scans were performed after background subtraction and the results averaged. The measurement data was later normalised to compensate for concentration eﬀects.

2.3.3 Karl-Fischer-titration (KFT)

Karl-Fischer-titration (KFT) was used for the determination of the amount of water in selected organic liquid samples. Samples were evaporated in a Metrohm 774 oven processor at 651 K. A volumetric method was performed with the titrant solution Hydranal Composit 5. The end point of the titration was marked with the bipotentiometric method.

19 2.3. Analysis

2.3.4 Photometry Photometric spectroscopy (UV-Vis) was applied for the quantification of phenolic compounds in aqueous samples. Applying a standardised test purchased from Hach Lange the sample is mixed with an oxidising agent. All ortho- and meta substituted phenols will form coloured complexes with 4- aminoantipyrine. The absorption of the light of a characteristic wavelength 510 nm is measured in a calibrated photometer. Usually this method is utilised as proof of evidence for the existence of phenolics in a sample. For the quantification of single phenolic compounds, carried out for determining the distribution in water and ethyl acetate (see Appendix B.2), each phenolic substance had to be calibrated individually. The quantification of a phenolic substance from a solution containing different monoaromatic substances is thus not possible.

CHNS analysis CHNS analysis was applied on the feedstock lignin and on samples from the liquid and solid product phase in order to determine the elemental composition of a sample. The utilised apparatus was a Vario El III by the company Elementar. The percentage ofC,H,N andS, which refer to the dry matter excluding the moisture, were directly determined with the Dumas-method. The residual percentage was suggested to beO and ash. The amount of the ash was measured by gravimetric analysis of the residue after heating the sample from 298 K to 1273 K in 4 h and subsequently maintaining the temperature of 1273 K for 2 h

20 Chapter 3

Thermodynamic Studies

3.1 Introduction

An important question for the evaluation of the experimental results is whether the reactions taking place within the thermal degradation of lignin reach a thermodynamic equilibrium within the predetermined residence time. The lignin which is placed into a batch reactor together with a solvent is isolated from its surroundings, and thus sooner or later thermodynamic equilibrium is reached. The knowledge about the thermodynamic equilibrium provides the possibility to predict the tendencies for the experimental results at increasing residence times and temperatures. The latter has to be treated with care since the reaction rate increases with the temperature and thus the time from the start of the reaction until the state of equilibrium shortens. In addition, the temperature influences the thermodynamic equilibrium itself. Hence, the evaluation of experimental results requires at least a basic knowledge about the involved reactions and the composition in the thermodynamic equilibrium. The composition of the product mixture after infinite residence time is compared with the experimental results at very long residence times. For the calculation of the composition in the thermodynamic equilibrium the program Aspen Plus V7.2 was employed. Aspen Plus uses different methods for the estimation of the thermodynamic equilibrium. Two of those were chosen for the present work. The group contribution method Unifac (Dort- mund) developed by Fredenslund, Jones, and Prausnitz [FJP75] is based on a GE-model [Pfe04]. The group contribution method describes any complex molecule as a combination of structural groups. This is highly advantageous

21 3.2. Input parameters when dealing with complex organic material. The excess Gibbs enthalpy as well as activity coefficient and fugacity coefficient were calculated by the model which uses the HYSYS-database included in Aspen Plus containing in excess of 1500 components and over 16000 fitted binaries. It is problematic for the Unifac (Dortmund) model to deal with supercritical components which is done with employing Henry`s law [Asp10]. Unifac (Dortmund) has become very popular due to its large range of applicability and the reliable results predicted for vapour-liquid equilibria as well as for solid-liquid equilibria over a wide temperature range [Gme+02]. Especially char formation may be a challenge for other models, such as the equation of state (EoS) developed by Peng-Robinson, which was developed for gas phase reactions and would thus need further revision if applied for the calculation of the chemical potential of solid reaction components [CF11]. However, for further comparison, the Peng-Robinson-EoS was as well employed to calculate the thermodynamic equilibrium using the minimum search of the Gibbs enthalpy of Aspen Plus since it was especially developed to model thermodynamic equilibrium in the range of the critical point. The literature version of the alpha function and mixing rules were used [PR76; Asp10]. The thermal degradation of lignin in two different solvents that are water an ethanol (EtOH) was evaluated. The product composition in the equilibrium state calculated with both methods, Unifac and Peng-Robinson-EoS were compared. Furthermore the comparison of the data obtained from the Aspen-calculation with experimental data was intended. Few data is available in literature picturing to the liquid-liquid equilibrium (LLE) or the gas-liquid equilibrium (GLE) for mixtures of aromatics in water or ethanol.

3.2 Input parameters

Model substances had to be defined as input material since lignin cannot be sufficiently characterised for the calculation in Aspen due to its inhomogeneous character. Experimental results (experiment (Exp.) 90, 158) thus serve as input parameters for the calculation. Input parameters for the calculation of the thermodynamic equilibrium were chosen according to experimental results at 120 min residence time. After 120 min an entire degradation of the lignin molecule into smaller fractions can be assumed (see Chapter6). The fraction of depolymerised lignin, which is soluble in ethanol or water, but could not be quantified byGC was described by the chemical compound bisphenol-A. In Table 3.1 all input components considered for the calculation with both methods Unifac and Peng-Robinson-EoS are listed. Table 3.1 includes the initial concentrations of the input components and, whether each

22 Chapter 3. Thermodynamic Studies component is to be found in the solid, liquid or gas phase in the equilibrium state. Temperature and pressure for the model calculation were chosen according to the experimental data (T = 633 K, p = 30 MPa). For the calculation of the product composition in the thermodynamic equilibrium additional components are expected; that are propane and propylene in the gas phase, benzene, cyclohexane, cyclohexanone, cyclohexanol and methanol in the liquid phase, anthracene, naphthalene and carbon in the solid phase. For easier comparison with experimental results the initial concentrations are displayed as initial yields yi,0. The input mass of lignin corresponds to the sum of all input masses except the input mass of the solvent, that is either water or ethanol.

Table 3.1: Initial yields yi,0 for the Aspen-calculation of thermodynamic equilibrium at 633 K, derived from the experimental results at 120 min in water (Exp. 90) and ethanol (Exp. 158, see Appendix E.1).

Compound Phase Water Ethanol mg/glignin

CO2 gas 166.81 136.49 CO gas 1.29 0

CH4 gas 7.56 15.13 H2 gas 0.01 0.1 C2H6 gas 0 29.66 C2H4 gas 0 20.89 C4H10 gas 0 2.09 H2O liquid 7577.62 197.02 Phenol liquid 1.97 1.39 Catechol liquid 25.76 1.21 Guaiacol liquid 3.33 2.7 Ethanol liquid 0 5982.54 Bisphenol-A solid 793.27 593.32

3.3 Results

The results of the minimum search of the Gibbs-enthalpy conducted by Aspen employing both the Unifac-method and Peng-Robinson-EoS for the thermal degradation of lignin in water and ethanol are displayed in Ta- ble 3.2. Products which, although considered by the model, yielded less −4 than 10 mg/glignin were not listed. The results obtained with the Unifac-

23 3.3. Results method and the Peng-Robinson-EoS are in good agreement to each other. Both methods suggest that in water all participating organic compounds are transformed into gaseous products, mainly CO2 and CH4, also smaller amounts of H2 and CO, but only traces of C2 to C4 hydrocarbon gases. On the contrary for the degradation in ethanol both methods suggest high yields of hydrocarbon gases, CH4 being the most dominant product. The percentage of hydrocarbon gases and CO is more than 100 times higher in ethanol than in water. The yield of CO2 is three times higher in ethanol than in water. Only the amount of H2 is higher in water according to the calculations. Furthermore, results shows that at given conditions (633 K and 30 MPa) water is more stable than ethanol. The initial ethanol disappears completely while more than 91 wt% of the initial amount of water is conserved. Phenolic components, that are phenol, catechol and guaiacol are practically absent in the product mixture. Solid carbon was neither calculated for hydrothermal nor for solvothermal degradation in ethanol by none of the two methods. In ethanol signiﬁcant yields of anthracene and naphthalene in the solid product phase are to be found in the equilibrium state according to the calculations

Table 3.2: Equilibrium yields yi of thermal lignin degradation in water and ethanol calculated with Unifac and Peng-Robinson-EoS at 633 K and 30 MPa, input concentrations are listed in Table 3.1. Compound yi in water yi in ethanol Unifac Peng-Robinson Unifac Peng-Robinson mg/glignin mg/glignin mg/glignin mg/glignin

CO2 1126.70 1125.84 3328.35 3328.52 CO 0.19 0.20 22.21 21.98

CH4 521.59 521.90 3339.67 3342.14 H2 1.08 0.93 0.12 0.12 C2H6 0.06 0.07 27.04 26.26 C2H4 0.00 0.00 0.01 0.01 C3H8 0.00 0.00 1.31 1.24 C3H6 0.00 0.00 0.01 0.01 C4H10 0.00 0.00 0.05 0.05 H2O 6927.98 6928.69 7.35 7.35 Anthracene 0.00 0.00 228.43 236.04 Naphthalene 0.00 0.00 52.61 61.22

The results obtained by calculation with Aspen were compared with experimental results from thermal degradation of lignin in water (Exp. 148) and ethanol (Exp. 152) at 18 h residence time. Figure 3.1 shows that after

24 Chapter 3. Thermodynamic Studies

1 0 0 1 0 0 a ) b ) ) % ) l o % t m ( w

(

n n o i o 5 0 t 5 0 i i t

E r r o r s c o a G a s p r f

L i q u i d m 2 s o s

S o l i d c C H

a 4

M C O a 2 G 0 0 U n i f a c P e n g - R o b i n s o n E x p W a t e r U n i f a c P e n g - R o b i n s o n E x p W a t e r

Figure 3.1: Thermodynamic equilibrium calculated by Aspen with Unifac- method and Peng-Robinson-EoS compared with results from Exp. 148 a) mass fraction of of product phases and b) composition of the gas phase; τ = 1080 min; spruce-lignin/water = 132 g/L; T = 633 K.

1 0 0 1 0 0 a ) b ) ) % l ) o % t m ( w C - C ( 2 4

n n o

i C H t

o 4

5 0 i 5 0 i t s

c C O

o 2

a E r r o r p r f

G a s m s o

s L i q u i d C a

S o l i d s M a G 0 0 U n i f a c P e n g - R o b i n s o n E x p E t O H U n i f a c P e n g - R o b i n s o n E x p E t O H

Figure 3.2: Thermodynamic equilibrium calculated by Aspen with Unifac- method and Peng-Robinson-EoS compared with results from Exp. 152 a) mass fraction of of product phases and b) composition of the gas phase; τ = 1080 min; spruce-lignin/ethanol = 132 g/L; T = 633 K.

25 3.4. Discussion

18 h residence time in water the amount of gas is signiﬁcantly lower than the calculation results. In addition, the solid residue recovered from the reactor after the experiment was 4.1 wt%. However, calculation results did not show any solid residue. The calculated and the experimentally determined amount of liquid are similar. The gas recovered from the experiment comprises approximately 11 and 89 mol% of methane and CO2 respectively. H2 was not detected in the gas phase. In the calculated thermodynamic equilibrium, however, methane makes up more than 50 mol% of the gas and H2 up to 2 mol%. In fact the experimental results at 18 h do not agree with the calculated values. In respect to the thermal degradation in ethanol similar phenomena were observed. The amount of gas recovered from the experiment at 18 h was very low (22 %) compared to the amount of the calculated gas amount (96 %). The experimentally determined amount of solid residues (2.7 %) is less than the calculated amount of solids (4 %) which comprises exclusively anthracene and naphthalene (see 3.2a). The analysis of the gaseous product of Exp 152 shows similar to the experiment in water approximately 91 mol% of CO2. The other 9 mol% of the gas are hydrocarbons, mostly methane (6 %). However, according to the calculations, in the equilibrium state methane makes up more than 70 mol% of the gas phase. Only traces of other hydrocarbons and 27 mol% of CO2 make up the rest of the gaseous product.

3.4 Discussion

The results of the Gibbs-enthalpy minimum search employing the Unifac- model or the Peng-Robinson-EoS shows that at given conditions (633 K, 30 MPa) the major part of the organic input material is gasified. The ab- sence of a liquid phase and the composition of the gaseous phase in the equilibrium state (see Figure 3.2) suggest that ethanol is completely transformed into hydrocarbons and CO2. The fact that the solvent is entirely or partially consumed by the occurring reactions negatively affects the technical feasibility of the lignin degradation in ethanol since it renders a recircula- tion and reutilisation of the solvent impossible. The solid product phase resulting from the calculation of the equilibrium state in ethanol makes up 5 wt% of the total product mass and about 30 wt% of the initial lignin mass. It is exclusively composed of aromates, that are anthracene and naphthalene. Hence, the solid phase comprises products from repolymerisation or recombination of aromatic intermediates, e.g. phenolics. Phenolics are thus not stable at the determined conditions. They are either gasified or polymerised in ethanol. In water the phenolic products are entirely gasified,

26 Chapter 3. Thermodynamic Studies probably since the presence of near critical water (NCW) and supercritical water (SCW) facilitates the degradation of the aromatic ring. Experimental results show that after 18 h 70 wt% of the initial organic liquid, mainly comprising ethanol, were recovered. This indicates at first that after 18 h the thermodynamic equilibrium is still not reached and sec- ondly suggests a rather slow gasification of ethanol. However, the calculation of the thermodynamic equilibrium and the associated assumptions described here have to be critically evaluated and eventually need improvement. The calculation of the build-up of char is challenging, and can be improved as described by Castello and Fiori [CF11]. The use of model components such as bisphenol-A is a good first approximation, but needs further revision. How- ever, here will always be a source of potential errors due to the heterogeneity of lignin.

27 3.4. Discussion

28 Chapter 4

Inﬂuence of solvent on thermal lignin degradation

4.1 Introduction

In this chapter the solvothermal depolymerisation of lignin in ethanol and the hydrothermal degradation in water are compared. The main focus lies on the formation and decomposition of selected phenolic components, as well as the formation of char and gas. For this purpose various experiments were carried out utilising diﬀerent approaches of hydrogen donation, that are elemental hydrogen in combination with a Pd-catalyst and in situ hydrogen formation via degradation of formic acid. All experiments were carried out in 5 mlMA2. For this reactor setup no measurement of the pressure was not applicable. The lignin used for this study was enzymatic hydrolysis lignin from spruce wood. Barth and Kleinert have done intensive research on the lignin solvolysis in ethanol (EtOH) using formic acid as hydrogen donor. They found a large variety of aliphatic and aromatic products comprising few oxygenated components [KB08b], almost no solid residue (< 5 wt%), a high H/C-ratio and a low O/C-ratio[Gel+08; KB08a; Kle+11].

4.2 Experimental Results

Experimental results of the lignin degradation in water and ethanol without hydrogen donor are explored. Experiments were carried out in aMA2 with 5 mL inner volume. The reactors were cooled with iced water after the

29 4.2. Experimental Results reaction. Figure 4.1 shows the results obtained from degradation of spruce lignin in ethanol and H2O at T = 633 K. The development of the solid yield over the residence time is similar for both solvents. After a ﬁrst decline, the yield of solid product slowly approaches a threshold value. This threshold is slightly higher for experiments in water (approximately 0.34 g/glignin) than in ethanol (approximately 0.27 g/glignin). a ) G a s b ) 1 0 0 S o l i d r e s i d u e 0 . 6 ) %

l 8 0

o C 4 - G a s e s ) V n ( i E t O H

n P r o p a n e g n i l 6 0 o P r o p e n e g

0 . 4 i / t i

g E t h a n e s (

o E t h e n e d p l 4 0 C O e m

i H O

2 o

Y C O 0 . 2 c 2

s 2 0 M e t h a n e a G

0 . 0 0 0 2 0 0 4 0 0 W a t e r E t O H R e s i d e n c e t i m e ( m i n )

Figure 4.1: Results from thermal degradation of spruce lignin in ethanol (EtOH) or water (H2O) at T = 633 K; spruce-lignin/solvent = 132 g/L; a) yield of gas and solid residue in EtOH (solid symbols, Exp. 153-161) and H2O (open symbols, Exp. 85-93); b) Gas composition for τ = 480 min

The development of the gas yield over the residence time is similar for both solvents until 100 min residence time. It is increasing with the residence time. The relatively low value at 60 min is considered to be an outlier. However, at residence times longer than 100 min the gas yield of ethanol experiments increases up to almost 0.7 g/glignin whereas the gas yield of water experiments stays constant at less than 0.2 g/glignin. TheGC-analysis of the gas reveals that the gas recovered from water experiments comprises mostly CO2 (90 vol%) and CH4 (9 vol%) and only 1 vol% of other hydrocarbon gases. Gas recovered from ethanol experiments on the contrary comprises more than 50 vol% hydrocarbon gases, mainly C2-gases, that are ethane (37 vol%) and ethene (6 vol%), methane (11 vol%) and also about 9 vol% of C3-C4-oleﬁnes. Only about 36 vol% of the recovered gas is CO2. Figure 4.2 demonstrates the FT-IR-spectra of the solid residue removed from the batch-autoclave after diﬀerent residence times are compared to the

30 Chapter 4. Inﬂuence of solvent on thermal lignin degradation

a ) b )

) 4 8 0 m i n 4 8 0 m i n . m r 6 0 m i n 6 0 m i n o n ( 3 0 m i n 3 0 m i n e c n

t 1 5 m i n

t 1 5 m i n i m

s 0 m i n 0 m i n n

a 1 5 1 1 1 5 1 1 r 1 2 7 0 1 2 7 0 1 4 2 9 1 4 2 9 1 1 6 0 T 1 1 6 0 1 1 1 2 1 0 3 3 1 1 1 2 1 0 5 8 1 0 3 3 1 0 5 8 1 6 0 0 1 4 0 0 1 2 0 0 1 0 0 0 1 6 0 0 1 4 0 0 1 2 0 0 1 0 0 0 W a v e n u m b e r ( c m - 1 ) W a v e n u m b e r ( c m - 1 )

Figure 4.2: FT-IR-analysis of spruce-lignin and the solid residue after 15, 30, 60 and 480 min residence time in a 5mLMA2 with a) ethanol (Exp. 153, 154, 156, 161) and b) H2O(Exp. 85, 86, 88, 93); lignin/solvent = 132 g/L; T = 633 K

FT-IR-spectrum of spruce-lignin. The solid samples were recovered from the corresponding experiments described above without addition of catalyst or hydrogenation agent. The displayed curve is reduced to the region between 900 and 1700 cm−1 since here most of the vibrational bands are visible which are related to the aromatic ring (also compare Table 6.1). The curves are normed and thus only a qualitative comparison is valid. It is clearly visible that most aromatic vibrational bands, e.g. at 1511 cm−1, 1429 cm−1 and 1058 cm−1, disappear towards residence times above 60 min. Beyond 60 min only minor changes of the spectrum are visible. At increasing residence time all peaks disappear and the curve approximates a horizontal line. The disappearance of the aromatic ring vibrational bands, e.g. at 1058 cm−1 is visibly faster in water than in ethanol. Figure 4.3 shows the chromatogram of the liquid products from experiments of lignin solvolysis in ethanol and in water. The hydrogenation was achieved by adding formic acid. The compounds corresponding to the diﬀer- ent peaks are listed in Table 4.1. The chromatogram shows the relative abundance of the substances present in the liquid product in case of ethanol experiments. Since the analysis of the aqueous liquid product was not possible in case of water experiments, an extraction of the phenolics into an organic extraction phase, that is ethyl acetate, was carried out (see Appendix B.2).

31 4.2. Experimental Results

1 . 0 E t O H 6 9 1 9 H O 8 e 2 c n a

d 4 n

u 1 4 b A 0 . 5 7 e

v 1 1

i 1 0

t 1 3

a 3 l 1 5 e 1 6 1 2 R 1 2 1 7 1 8 5 0 . 0 1 0 1 5 2 0 2 5 3 0 R e t e n t i o n T i m e ( m i n )

Figure 4.3:GC-FID-chromatogram of liquid product (EtOH, Exp. 143) and organic phase after extraction of aqueous liquid product (H2O, Exp. 142) after thermal treatment of spruce-lignin; lignin/solvent = 132 g/L; FA/solvent = 107g/L; T = 633 K, τ = 60 min

32 Chapter 4. Inﬂuence of solvent on thermal lignin degradation

Hence, the chromatogram in Figure 4.3 labelled with »H2O« shows the analysis of the extraction phase. Due to the specific coefficient (fd,i), which indicates the transfer ratio of each phenolic compound from the aqueous to the organic phase, the peak area is influenced by the extraction which has to be considered for the interpretation of the chromatograms. The chromatograms reveal that after the thermal degradation of spruce-lignin in water catechol is the predominant phenolic product. Solvolysis in ethanol on the other hand yields higher amounts of ethyl substituted phenols and methoxyphenols than hydrothermal degradation of lignin. In Figure 4.3 the peak of 4- ethylphenol and 4-ethylguaiacol in the H2O curve is practically not visible. Furthermore, the peaks of syringol and its alkylated derivatives are clearly more dominant in the liquid sample after solvolysis in ethanol. Among the non phenolic compounds present in the liquid products, cyclopentanone and 2-methylcyclopentanone are more abundant after hydrothermal treatment. Ethanol yields aliphatic compounds such as 1,1-diethoxyethane. Generally, the number of different phenolic compounds in hydrothermal treatment is lower.

Table 4.1: Identiﬁed products with corresponding retention time in theGC- FID. Compound # Name Molecular Retention class formula time (min)

Methoxyphenols 4 guaiacol C7H8O2 18.36 7 4-methylguaiacol C8H10O2 21.58 9 4-ethylguaiacol C9H12O2 24.03 11 syringol C8H10O3 25.63 13 4-methylsyringol C9H12O3 28.20 14 4-ethylsyringol C10H14O3 30.13 15 4-propylsyringol C11H16O3 32.06 Catechols 8 catechol C6H6O2 22.05 10 4-methylcatechol C7H8O2 24.48 12 4-ethylcatechol C8H10O2 26.85 Phenols 1 phenol C6H6O 15.25 2 2-methylphenol C7H8O 17.61 3 4-methylphenol C7H8O 18.27 5 2-ethylphenol C8H10O 20.18 6 4-ethylphenol C8H10O 21.02 Others 16 1,1-diethoxyethane C6H14O 6.22 17 cyclopentanone C5H8O 7.47 18 2-methylcyclopentanone C7H8O 9.47 19 pentadecane C8H10O 30.30

33 4.3. Discussion

1 6 4 0 P h e n o l ) n

i G u a i a c o l n ) g i n l 1 2 C a t e c h o l 3 0 i n g g i / l g g / m g (

8 2 0 m a (

u l o G

4 1 0 h c &

e t e a h

P 0 0 C 5 6 0 6 0 0 6 4 0 6 8 0 7 2 0 T ( K )

Figure 4.4: Yield of diﬀerent phenolics from lignin solvolysis in ethanol (solid symbols, Exp 126-133) and H2O (open symbols, Exp 134-141) at diﬀerent temperatures; spruce-lignin/solvent = 80 g/L; 8 mg Pd-catalyst; 1 MPa H2 initial pressure; τ = 60 min

Experiments of thermal degradation of spruce-lignin with elemental hydrogen combined with a Pd-catalyst conﬁrm that in the presence of water the amount of catechols increases signiﬁcantly. Above 673 K the hydrothermal treatment of spruce-lignin yields up to 38 mg/glignin catechol. At any tested temperature the yield of catechol after thermal degradation of spruce-lignin in ethanol was at most 5 mg/glignin (see Figure 4.4). Furthermore, the results of experiments with Pd-catalyst show that the sensitivity of the yields of all contemplated compounds, that are guaiacol, catechol and phenol, in respect to changes of the temperature at 60 min residence time is higher for hydrothermal experiments than for ethanol solvolysis. This becomes especially visible around the critical temperature of water (Tc,W = 647 K).

4.3 Discussion

The results presented in this chapter show that water is an eﬀective solvent in respect to the recovery of phenolics from lignin and selective in respect to the yield of catechols. FT-IR-analysis of the solid residue reveals that spruce-lignin degrades faster in water than in ethanol. In respect to the FT- IR-spectrometry of the solid residue, typical aromatic vibrational bands, e.g.

34 Chapter 4. Influence of solvent on thermal lignin degradation at 1058 cm−1, disappear after less than 15 min in water whereas in ethanol these bands are still visible after 30 min (see Figure 4.2). This is probably due to different reaction mechanisms in both solvents. In a hydrothermal environment hydrolysis favours the cleavage of C-O bonds. This is also discussed more detailed in Chapter 6.3. Compared to the solvolysis in ethanol, hydrothermal treatment shows a narrower products spectrum of phenolic products. Especially ethylated phenols and methoxyphenols are hardly detectable in the aqueous liquid product. The abundance of 4-ethylphenol and 4-ethylguaiacol in the presence of ethanol indicates that the solvent itself is involved in alkylation reactions. The formation of 1,1-diethoxyethane as well as butane and propane in ethanol is also explicable by the combination of smaller olefines or ethanol. Hence, the high reactivity of ethanol in comparison to water leads to a broader product spectrum. Lignin degradation in water, which is basically inert at the determined conditions, on the contrary yields relatively high amounts of catechol and its derivatives. This is an advantage in respect to the economical feasibility since a decreased number of substances in the product mixture and an increased selectivity in respect to few phenolic products facilitates the separation of the products from waste and solvent in the process. Cyclopentanone and 2-methylcyclopentanone, which are present in the liquid sample after hydrothermal degradation of lignin, most probably originate from the residue cellulose in the spruce-lignin. Solvolysis in ethanol yields higher amounts of gas, probably due to the gasification of ethanol since increased concentrations of ethane and ethene are observed. The lower amount of char in ethanol can be explained by the increased solubility of organic components, such as aromatic oligomers produced by e.g. repolymerisation of phenolic monomers [McM+04], in ethanol at room temperature.

35 4.3. Discussion

36 Chapter 5

Inﬂuence of catalysts on thermal lignin degradation

5.1 Introduction

In this chapter experimental results concerning the influence of heterogeneous catalysts on lignin degradation are presented. Heterogeneous catalysts have a high potential to promote hydrogenation reactions, e.g. the hydrogenation of phenolic intermediates. Homogeneous catalysts, being less limited by transport phenomena, are more likely to interact with the lignin- macromolecule and thus favour the cleavage of the bonds between aromatic monomers [Beh+06]. The focus of the investigation lies on heterogeneous catalysts comprising different metals. The catalysts were selected following the recommendations of other research documentations (see Chapter 1.2.5). Rh, Pt, Pd and Ni already proved to catalyse the hydrodeoxygenation of phenolics. Pd and Ni were also used in hydrothermal conditions [Zha+10a]. cobalt (Co) is a standard hydrogenation catalyst for the hydrodeoxygenation of pyrolysis-oil [Gut+09], however, has not been tested in a hydrothermal environment. All selected heterogeneous catalysts were from typical hydrogenation catalysts of the eighth subgroup of the periodic table of the ele- ments. A screening of the effects of homogeneous catalysts, that are acidic and basic solutions, on lignin degradation can be found in Appendix D.1.

37 5.2. Screening of heterogeneous catalysts

a ) b ) N o C a t . 2 4 2 4 R A N E Y - N i 2 0 2 0 R h

) ) C o n n i i

n 1 6 n 1 6 P t g g i i l l

g g P d C l / /

g 1 2 g 1 2 A l - N i m m ( ( 8 8 d d l l e e i 4 i 4 Y Y

0 0 H H o l s H H o l s h O h O c h h O h O c h P y - P t e P y - P t e x C a x C a t h o t h o M e M e

Figure 5.1: Eﬀect of diﬀerent heterogeneous catalysts on the yield of phenols, methoxyphenols and catechols from hydrothermal lignin degradation (Exp. 8-20 and 27-31), spruce-lignin/water = 133 g/L; T = 633 K; τ = 30 min; 0.10 g/glignin Raney-Ni, 0.08 g/glignin Rh on act. C, 0.08 g/glignin Pt on act. C, 0.07 g/glignin Co, 0.08 g/glignin PdCl2, 0.13 g/glignin Al-Ni; reactor pressurised with a) N2 (5 MPa initial pressure) and b) H2 (5 MPa initial pressure).

5.2 Screening of heterogeneous catalysts

It was shown in the past that heterogeneous catalysts have a high potential to promote hydrodeoxygenation of methoxyphenols which are present in bio- oils [16, 17]. Elemental hydrogen is supposed to be absorbed on the surface of the hydrogenation catalyst in order to be transformed into reactive hydrogen radicals. The latter react with the organic lignin fragments, e.g. phenolics. It is suggested that in situ produced hydrogen from the lignin degradation takes part in the hydrogenation of phenolics in a similar way. An important question in this context is whether the addition elemental hydrogen is indispensable for the catalytic hydrogenation of the phenolics or whether the in situ production of hydrogen from lignin degradation is suﬃ- cient. In order to answer this question experimental results of catalytic lignin degradation under H2 atmosphere were compared with results obtained under N2 atmosphere. The gas, H2 or N2, was pressed with an initial pressure of 5 MPa into the 5 mLMA2 which had been loaded with lignin, water and catalyst. Figure 5.1a shows the yields of phenols, methoxyphenols and catechols after hydrothermal catalytic degradation of spruce-lignin under N2

38 Chapter 5. Influence of catalysts on thermal lignin degradation atmosphere, Figure 5.1b shows the results from experiments with additional H2. The results show that the influence of additional elemental H2 is negligible since the yields of all quantified phenolics are similar and within the range of uncertainty. The results of a screening of different heterogeneous catalysts, namely Raney-Ni, rhodium on activated carbon (Rh on act.C), platinum on activated carbon (Pt on act.C), cobalt powder (Co), palladium-chloride (PdCl2) and aluminum-nickel (aluminium (Al)-Ni) is shown in 5.1. The experimental results reveal that among the investigated catalysts Raney-Ni and PdCl2 in the used concentrations have a major influence on the yield of the phenolic products. Phenol, o-cresol and p-cresol showed similar tendencies and were thus summarised within the (lump)-component phenols (PhOH). In the same way guaiacol and 4-methylguaiacol were summarised within the component methoxyphenols (methoxyPhOH) and catechol and 4-methylcatechol within the component catechols. The thermal degradation of spruce-lignin with Raney-Ni shows a sixfold increase of the yield of phenols from 1 to maximum 6 mg/glignin compared to experiments without catalyst. The application of PdCl2 on the thermal lignin degradation reduces the yield of methoxyphenols from more than 5 mg/glignin to less than 2 mg/glignin and increases the yield of catechols from about 5 mg/glignin to about 20 mg/glignin.

5.3 Raney-Nickel

The screening experiments presented in the preceding section show the effect of Raney-Ni on the formation of phenols. The latter are the phenolic products from lignin degradation with the largest variety of applications in the chemical industry. After evaluation of the catalyst screening experiments a more detailed study of the effect of Raney-Ni on the hydrodeoxygenation of intermediate phenolic products seems to be valuable. In a series of experiments, the input mass of Raney-Ni is varied at a constant temperature (633 K) and residence time (30 min). For a better representation of the data the phenolic products were grouped as described in the preceding section. The results are demonstrated in Figure 5.2. They reveal that with increasing input amount of Raney-Ni the yield of methoxyPhOH and catechols decreases and the yield of PhOH increases. At very high input concentrations of Raney-Ni (>250 mg/glignin) the yield of phenol decreases. In order to find out more about the reaction mechanisms which lead to an increased yield of phenols, additional experiments with intermediate substances were carried out. Catechol was diluted in water and placed together with 55 mg Raney-Ni into aMA2. The autoclave was pressurised with 1 MPa

39 5.3. Raney-Nickel

1 0 P h O H )

n M e t h o x y P h O H i n g

i 8 l C a t e c h o l s g /

g 6 m (

l 4 e i

Y 2

0 0 1 0 0 2 0 0 5 0 0

R a n e y - N i ( m g / g l i g n i n )

Figure 5.2: Yield of PhOH, methoxyPhOH and catechols at diﬀerent loadings of Raney-Ni (Exp 10-11 and 21-31); spruce-lignin/water = 133 g/L; T =633 K; τ =30 min.

1 0 0 0 n o t d e t e c t e d n o t r e a c t e d ) l

o C O h 2 c e t

a C H 4 C g

/ 5 0 0 C y c l o h e x a n o n e g C y c l o h e x a n o l m (

p - C r e s o l d l P h e n o l e i Y 0 5 2 3 5 7 3 6 0 3 6 2 3 T e m p e r a t u r e ( K )

Figure 5.3: Product composition after hydrogenation of 10 g/L catechol in water at diﬀerent temperatures with 55 mg(dry matter) Raney-Ni (Exp 36- 39); 1 MPa initial pressure H2; τ = 60 min.

40 Chapter 5. Inﬂuence of catalysts on thermal lignin degradation

H2 in order to guarantee a hydrogenation of catechol. The in-situ production of H2 might as well occur but is supposed to be inferior for the thermal decomposition of catechol compared to the thermal degradation of lignin. The mixture of reactants was exposed to diﬀerent temperatures from 523 K to 623 K with a residence time of 60 min. The results displayed in Figure 5.3 reveal that the highest yield of phenol of 229 g/gcatechol is obtained at 523 K. The phenol yield decreases with increasing temperature. However, the yield of gaseous products increases up to 296 g/gcatechol CO2 and 457 g/gcatechol CH4 at 623 K. Other important byproducts are cyclohexanol and cyclohexanone. The inﬂuence of temperature and reaction time on the Raney-Ni catalysed reaction from catechol to phenol is described in more detail in Appendix D.2, where furthermore the reaction pathways within catalysed hydrothermal catechol decomposition and their kinetics are discussed.

1 0 0 5 7 3 K 8 0 ) 5 2 3 K %

( 4 7 3 K

6 0 y t i v

i 4 0 t c e l 2 0 e S 0 2 0 4 0 6 0 8 0 1 0 0 C o n v e r s i o n ( % )

Figure 5.4: Selectivity of phenol over catechol conversion after hydrothermal catechol decomposition catalysed with Raney-Ni (Exp. 103-105, 114-118 and 122-125); input: 5 ml catechol/water-solution (5 g/L); diﬀerent temperatures and residence times.

Numerous experiments concerning the hydrothermal decomposition of catechol catalysed by Raney-Ni were carried out in 10 mLMA2. It was observed that the selectivity of phenol increases with the catechol conversion and reaches its maximum of about 90 % with the corresponding conversion of catechol of 88 %. At conversions above 90 % the selectivity of phenol rapidly decreases (see Figure 5.4). The graph shows a coefficient of determination R2 of 0.44 for the linear fit of the selectivity at a conversion lower than 80 % and a R2 of 0.85 for the linear fit at a conversion higher than

41 5.4. Discussion

80 %. This tendency is independent from the reaction temperature. Else- where the results of analogue experiments with guaiacol are documented [For+12], which show that phenol does not play a major role as product from the Raney-Ni catalysed hydrothermal decomposition of guaiacol. More important products are cyclohexanol and cyclohexanone.

5.4 Discussion

Heterogeneous catalysts proved to be effective on the hydrodeoxygenation of phenolic intermediates even in water. In addition, the in-situ production of hydrogen from hydrothermal lignin degradation is sufficient for the catalytic hydrogenation. PdCl2 was observed to yield high amounts of catechols and low amounts of methoxyphenols from the hydrothermal lignin degradation. It can thus be assumed that it catalyses the reaction from methoxyphenols to catechols (see Figure 5.1). However, this should be proven by an experimental study of the hydrothermal decomposition of methoxyphenols in the presence of PdCl2. The studies of the hydrodeoxygenation of aromatic substances in a hydrothermal environment carried out by Zhao and Lercher showed that Pd catalyses selectively the hydrogenation of the aromatic ring [Zha+1]. Ni on the contrary promotes »hydrogenolysis to cleave the C-O bonds, followed by removal of the oxygen atoms anchored to the aromatic moieties by sequential hydrogenation and dehydration reactions« [ZL12]. These findings agree with the results of experiments with Raney-Ni in the present work. Raney-Ni catalyses selectively the conversion from catechols to phenols, which is analogue to the cleavage of the C-O bond and the removal of an oxygen atom anchored to the aromatic ring. The maximum selectivity of phenol was obtained at 523 K. With increasing temperature and residence time cyclohexanol and cyclohexanone as well as gaseous products, that are CH4 and CO2, were observed after the catalytic hydrothermal decomposition of catechol with Raney-Ni. This indicates hydrogenation of the aromatic ring, as observed by Zhao and Lercher, but also gasification of the aromatic substances. Kruse et al. showed that the presence of Raney-Ni favours the gasification of aromatics in NCW and SCW[Kru+00; SKR04]. Probably this is also the reason for decreasing yields of phenol from spruce lignin with increasing input amounts of Raney-Ni beyond 250 mg/glignin. The yields and selectivities of phenol from catechol were compared with a study carried out by Wahyudiono, Sasaki, and Goto of the hydrothermal decomposition of catechol at temperatures between 643 and 663 K without catalyst [WSG09]. They documented a maximum selectivity of 20 %-25 % in respect to the molar input amount of catechol. The addition of Raney-Ni

42 Chapter 5. Influence of catalysts on thermal lignin degradation increases the phenol selectivities to 90 % at 88 % conversion of catechol at a significantly lower temperature of 523 K. Raney-Ni is thus a powerful catalyst for the hydrothermal conversion from catechol to phenol. The study of the influence of homogeneous catalysts on the hydrothermal lignin degradation (see Appendix D.1) indicates that basic catalysts, that are KOH and NaOH, increase the yield of catechols. HCL decreases the yield of methoxyphenols and increases the yield of catechols. These results agree with the assumption that both acetic and basic catalysts facilitate hydrolysis which favours the cleavage of C-O bonds e.g. in the methoxyl-group [TRR11; Zha+1]. The application of homogeneous catalysts is a promising way to increase the yield of phenolics from lignin. However, the results obtained here need to be complemented in order to be sufficient for stating clear tendencies.

43 5.4. Discussion

44 Chapter 6

Kinetic studies of lignin degradation

Kinetic models describing the depolymerisation of lignin are limited and are largely concentrated on global lump models describing the yields of gas, char and liquids. This chapter provides systematically obtained experimental data interpreted with a model which considers the kinetics of the crucial pathways (reactions) respecting the overall mass balance at any time. The interpretation of the model and the kinetic data obtained helps understanding the reaction mechanisms and determining bottleneck reactions. The first attempt of modelling the kinetics of lignin degradation was based on data obtained from the thermal degradation of lignin from wheat straw in ethanol and formic acid in a batch-autoclave at 633 K (see Section 6.1). The same model was further employed to describe the experimental data obtained in a CSTR (see Section 6.2). Finally, on the basis of the existing model for ethanol and formic acid, a modified model was developed which was able to describe the hydrothermal degradation of lignin from spruce in a batch- reactor (see Section 6.3). The modelling of lignin degradation in systems which are different in respect to the solvent, water or ethanol, and the reactor, batch reactor or CSTR, and the comparison of the results provide the possibility to a more profound knowledge about the influence of different process parameters on lignin degradation.

45 6.1. Lignin depolymerisation in ethanol in a batch-reactor

6.1 Lignin depolymerisation in ethanol in a batch-reactor

A formal kinetic model describing the main reaction pathways yielding phenolics, char and gas from wheat straw lignin in a hydrogen enriched solvolysis reaction system using ethanol as solvent has been developed. 10 experiments (Exp. 44-53, see Appendix E.1) were conducted using a mixture of 0.33 g lignin, 0.27 g formic acid and 2.5 g ethanol. The reactions were carried out at a constant temperature of 633 K and residence times of 15, 30, 45, 60, 90, 120, 240, 360, 480, and 1180 min. For comparison with the CSTR (see Section 6.2) note that in a Batch reactor the residence time is equal to the experimental duration. For the experiments in a CSTR, however, both terms have to be distinguished. Replicate experiments were performed for 60, 120 and 240 min residence times to assess experimental uncertainty.

6.1.1 Main intermediates and products Global mass balance The products comprise three major phases in varying ratios: gas, solid and liquid phase. The recovered amounts of these three phases are dependent on the residence time, and are illustrated in a global mass balance diagram in Figure 6.1. The decomposition of formic acid at 633 K contributes largely to the initial high gas yields. Extended residence time shows a continued growth in the amount of gas and a reduction of the single liquid phase obtained. Minimum amounts of solid residue in reference to the amount of input lignin are reached at 90 min (1.2 wt%) and 240 min (1.0 wt%). Although pressures in the reactors could not be measured, comparable experiments in a larger batch reactor have shown reaction pressures between 29 MPa and 33 MPa [Gas+10]. Based on the decreased formic acid to ethanol loading ratio, the pressure in the presented experiments is expected to be reduced by 3-6 MPa.

Gas phase Analysis of the diﬀerent components comprising the gaseous phase, see Fig- ure 6.2, show that up to 1.9 wt% of CO is produced at low residence times. The yield, however, quickly decreases to be replaced by a rapid increase of CO2. The high value (1600 mg/glignin) of the CO2 yield at 240 min residence time is supposed to be an outlier. Although H2 is detected, it is only present in minor amounts of some 0.01 wt% and further diminishes with increased

46 Chapter 6. Kinetic studies of lignin degradation

1 0 0 % F A H O

5 0 % t e r r o r

E g a s l i q u i d s o l i d L i g 0 % 0 1 5 3 0 4 5 6 0 9 0 1 2 0 2 4 0 3 6 0 4 8 0 1 1 8 0 R e s i d e n c e t i m e ( m i n )

Figure 6.1: Liquid, gas and solid reaction products at varying residence times are compared with the input amounts of Lig, EtOH and FA at reaction time zero of the experimental series. All experiments (Exp. 44-53) were conducted at the same loading conditions: 0.33 g Lig, 0.27 g FA, and 2.5 g EtOH; T = 633 K.

1 6 0 0

1 2 0 0

8 0 0 (

4 0 0

0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 1 0 0 0 1 2 0 0 (

Figure 6.2: Individual yield of the gaseous product components, quantiﬁed onGC-FID andGC-TCD. The values for CO and H 2 have been multiplied by a factor of 10, respectively 10000, to enable a better visibility on the given scale; T = 633 K.

47 6.1. Lignin depolymerisation in ethanol in a batch-reactor

4 0 0 1 2 0 ) n i )

t r a n s i t i o n z o n e n n i g i n l g i g l / g g /

g G a s

2 0 0 6 0 m ( m S o l i d s ( s

d s i l a o G S

0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 1 0 0 0 1 2 0 0 R e s i d e n c e t i m e ( m i n )

Figure 6.3: Gas and solid residue yields for experiments carried out at varying residence times. A transition zone, indicating the change from largely lignin characteristic solids to char is indicated, utilising the results from FT-IR analysis; T = 633 K. residence time. Degradation of formic acid has been reported to occur via two different pathways, yielding either CO and H2O or CO2 and H2 [YS98]. The alcoholic solvent has been seen to influence the ratio of the two main degradation pathways of formic acid to yield CO2 and H2 or CO and H2O in a roughly 3:2 ratio [Kle+11]. The rapid consumption of H2 suggests a central function and fast incorporation in the depolymerisation of the poly- meric lignin structures. The continued increase in total amount of gaseous products and the formation of hydrocarbon gases (including CH4,C2H6, C2H4,C3H8,C3H6,C4H10) points towards the importance of gasification and cracking in later phases of the reaction process.

Solid Residue The solid residue was weighed after the reaction was terminated. Due to repolymerisation reactions causing an increase of solids after an initial decrease, additional analysis of the solids were required to allocate the transition point between a predominantly unreacted lignin-type structure and polyaromatic char (see Figure 6.3). To estimate the degree of depolymerisation and conversion of lignin, as well as the transition to coking, FT-IR spectra of the solid residue and a sample of the input lignin were taken. Figure 6.4 shows the FT-IR spectrum of the unreacted lignin with labeled characteristic absorption bands. The spectrum reﬂects the predominant aro-

48 Chapter 6. Kinetic studies of lignin degradation matic ring structures with hydroxyl and C-O moieties. The relatively identical intensities of the absorption bands at 1511 and 1603 cm−1 indicate an increased ratio of syringyl relative to guaiacyl moieties in the polymer [FW87, pp 151-154]. The largely abundant para-coumaryl moieties do not enable a characteristic band allocation. All allocated origins of bands and their alterations throughout residence time are summarised in Table 6.1.

3 . 0 ) d

e 1 1 8 0 m i n s

2 . i 5 l a 1 2 0 m i n m

2 . r 0 o n

( 9 0 m i n

1 . e 5 6 1 8

c 4 6 0

n 0 m i n 2 3 6 1 2 3 3 8 1 3 6 6 a

1 . t 0 t i 1 3 2 9 1 1 5 6 8 3 8 2 8 5 0 m

s 1 0 3 3 0 . 5 2 9 2 3

n 1 7 0 0 2 9 5 7 1 1 2 0 a 3 5 0 0 - 3 4 0 0 1 4 2 9

r 1 6 0 3 1 4 6 2 1 2 1 8 T 1 2 7 1 0 . 0 1 5 1 1 4 0 0 0 3 5 0 0 3 0 0 0 2 5 0 0 2 0 0 0 1 5 0 0 1 0 0 0 5 0 0 W a v e n u m b e r ( c m - 1 )

Figure 6.4: FT-IR spectra of unreacted lignin and of the collected solid residues at 90, 120 and 1180 min residence time. The largest alterations take place during the ﬁrst 120 min and are documented in Table 6.1; T = 633 K.

Large alterations can be observed in the FT-IR spectrum of the solid residues within the ﬁrst 120 min residence time (see Figure 6.4) for four example spectra taken from the lignin and solid residues from the reactors after 90, 120 and 1180 min. The most dominant alteration is the disappearance of hydroxyl functions, which is seen in the decrease of the broad -OH stretching band at 3500-3400 cm−1. Decreasing absorbance of aromatic ring vibrational bands (1511, 1429 cm−1), especially in combination with C=O stretching (1700, 1329, 1271, 1218, 1120, 1033 cm−1) are clearly visible, illustrating the decomposition to monoaromatic units, and their solubility in the oil. These bands play a role in estimating the transition period from a largely lignin-like structure to a polyaromatic char. Roy, Bag, and Sen quantiﬁed the content of lignin, using the aromatic ring vibrational band at 1511 cm−1 [RBS87]. Analysis of collected FT-IR data, shows the disappearance of this

49 6.1. Lignin depolymerisation in ethanol in a batch-reactor band between 90 and 120 min.

Liquid Phase The liquid single-phase recovered was of a dark brownish colour and of a pungent odour. The single phase remained stable, without any precipitation, phase separation or change of chemical composition as observed byGC analysis, also after several weeks of storage at 278 K. For analysis, the crude product was evaluated without further separation, thus retaining the original amount of remaining ethanol and process water. The water content was found to increase from 3.9 wt% at 30 min to 14.7 wt% at 1180 min residence time. One major focus of the kinetic model are the selected key components. For the monoaromatic compounds quantiﬁed in the liquid phase, the maximum error for the single replicates were found be 5.3 wt%, which is considered to be within an acceptable range, furthermore, consistent trends for the development over the residence time of all phenolic products are clearly visible in Figure 6.5.

6.1.2 Grouped products The focus of the liquid analysis is to evaluate the monomeric phenolics in terms of their yields as a function of residence time. For this, different phenolics were grouped according to their degree and type of oxygenation. Their time-dependent yields are given in Figure 6.5. 4-Ethyl substitution is the dominant substitution for all species. In addition, 2-ethyl and 2- and 4-methyl substituted phenolics were quantified and included in some of the compound classes. Figure 6.5a shows the methoxylated phenolics which an initial rapid increase until approximately 90 min is followed by a decrease. The by comparison more rapid decrease of syringol supports the previously observed conversion of syringol to guaiacol [Dor+99]. In general, all methoxylated phenolics, including syringol, show a large similarity in their time-dependent yield profile. A B-spline function based on averaged values of the single constituents helps to illustrate the overall trends. The profile suggests the single constituents to be primary products within the analysis focus, and subsequently the rapid decrease in concentration supports their identification as intermediates. According to several investigations conducted under pyrolysis and hydrolysis conditions [JK85; LK85], catechols are found to be products of the secondary decomposition of methoxylated phenolics. Figure 6.5b shows the

50 Chapter 6. Kinetic studies of lignin degradation

Table 6.1: Characteristic FT-IR absorptions and observed changes between the biomass feed and the solid residue samples Observed Litera- Origin[Her71; LI92; LPH97; Peak character- Peak alterations band ture Ye+12] istics in in cm−1 cm−1 unreacted lignin solid residue 3500- 3450- -OH stretching broad initial strong de- 3400 3400 crease 2957, 2940- -OH stretching in methyl very weak, in- strong decrease 2923, 2820 and methylene groups tense, weak at 2923 cm−1 2850 2361, weak constant de- 2338 crease 1700 1715- C=O stretching non- disappears after 1710 conjugated to the aromatic 30 min ring 1675- C=O stretching in conju- 1660 gated to the aromatic ring 1603 1605- Aromatic ring vibrations intense decreases 1600 1511 1515- Aromatic ring vibrations intense disappears after 1505 120 min 1462 1470- C-H deformations (asym- 1462 metric) 1429 1430- Aromatic ring vibrations disappears after 1425 30 min 1366 1370- C-H deformations (symmet- weak intensiﬁes 1365 ric) 1329 1330- Syringyl ring breathing with weak disappears after 1325 C-O stretching 45 min 1271 1270- Guaiacyl ring breathing with disappears after 1275 C-O stretching 45 min 1218 1220 Syringyl ring breathing weak disappears after 60 min 1156 1160 Aromatic C-H in plane de- weak formation, guaiacyl type 1120 1120 Aromatic C-H in plane de- weak formation, syringyl type 1033 1030 Aromatic C-H in plane deformation guaiacyl type and C-O deformation of pri- disappears after mary alcohol 45 min 917 weak 867 appears at 240 min 838 834 para-substituted aromatic disappears at group 90 min 810 appears at 240 min 618 appears at 90 min 460 appears at 90 min

51 6.1. Lignin depolymerisation in ethanol in a batch-reactor

) a ) G u a i a c o l b 5 C a t e c h o l 6 4 - M e t h y l g u a i a c o l 4 - E t h y l c a t e c h o l 4 4 - M e t h y l c a t e c h o l ) 5 4 - E t h y l g u a i a c o l ) n n i i C a t e c h o l s ( a v . ) n S y r i n g o l n g g i i l 4 M e t h o x y p h e n o l s ( a v . ) l 3 g g / /

g 3 g 2 m m

( 2 (

d d 1 l 1 l e e i i

Y 0 Y 0 0 1 0 0 2 0 0 3 0 0 4 0 0 1 0 0 0 1 2 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 1 0 0 0 1 2 0 0

c ) 5 R e s i d e n c e t i m e ( m i n )

4 )

n P h e n o l i n g

i 2 - M e t h y l p h e n o l l 3

g 4 - M e t h y l p h e n o l /

g 2 2 - E t h y l p h e n o l

m 4 - E t h y l p h e n o l (

d 1 P h e n o l s ( a v e r a g e d ) l

e 4 - E t h y l p h e n o l i

Y 0 0 1 0 0 2 0 0 3 0 0 4 0 0 1 0 0 0 1 2 0 0 R e s i d e n c e t i m e ( m i n )

Figure 6.5: Yields of key components for the different phenolic compound classes: a) methoxyphenols, b) catechols and c) phenols over residence time; T = 633 K. different quantified catechols as well as the averaged trend function. A some- what slower initial increase in comparison to the methoxylated species with lower yields is found, together with a slight decrease at over 90 min residence time. However, the two values for catechol at 240 and 1180 min differ largely from the values for the substituted species, breaking with the generally decreasing yields. These high values were treated as outliers for further analyses. The dominance of 4-ethyl substituted species is especially noticeable for 4-ethylphenol in Figure 6.5c. Indeed, the rapid yield increase of this compound can be compared more to the increase of the methoxyphenols than to other phenolic constituents. The slow decrease over time however suggests that it is a more stable product within the degradation and reaction pathways of the monomeric units. The other non-methoxylated phenolics show a very slow yield increase, which when averaged, remains at a roughly con-

52 Chapter 6. Kinetic studies of lignin degradation stant yield value after 240 min residence time. The initial strong increase of 4-ethylphenol suggests a different reaction pathway compared to the rest of the phenols. Demethanisation and deoxygenation reactions link the different compound classes (see Table 4.1). However, the overall yield shows that a large proportion of the components is lost when increasing the residence time. This is due to char-formation or gasification processes.

6.1.3 Kinetic model development Discerning the rank of key reaction steps Delplot analysis was performed to confirm the indicative ranks of the phenolic species which had been grouped according to their compound classes as described above. Due to the uncertainty of the compositional transition from lignin to char, the necessary overall conversion rate for Delplot analysis was approximated by summing up all quantified monoaromatic species in the liquid. Full conversion is described by a maximum yield (indifferent as to the different classes involved) for all of these components. This is reached at 60 min residence time, after which the total yield declines, first slightly and then more rapidly.

a ) ) 1 0 0 b ) % ) (

2 0 2 % ( x

/ i x y / i

2 y

1 5 M e t h o x y P h O H n n o i o

i 4 - E t P h e s s r

r 1 0 1 0 P h O H e e v v

C a t e c h o l s n n o o C / C 5 / d l d l e i e i Y

Y 0 1 0 2 0 4 0 6 0 8 0 1 0 0 0 2 0 4 0 6 0 8 0 1 0 0 C o n v e r s i o n ( % ) C o n v e r s i o n ( % )

Figure 6.6: Methoxyphenols, phenols, 4-ethylphenol and catechols in dependence from the conversion, a) ﬁrst rank Delplot, b) second rank Delplot; T = 633 K

Figure 6.6a shows the ﬁrst rank Delplot with the three component lumps and 4-ethylphenol. The intercepts of the extrapolated trend lines towards zero conversion show that the methoxylated phenols and 4-ethylphenol are clearly ﬁnite, indicating them to be primary products. Clearly, considering the possible decomposition of syringol to form guaiacol, the latter can also

53 6.1. Lignin depolymerisation in ethanol in a batch-reactor be a secondary product. However, this reaction pathway was not considered for the present work. The extrapolated intercept of the y-axis, at the value zero, for the catechol lump shows it to be of a higher rank and thus at least a secondary product. The phenol lump intercepts the y-axis slightly under the value of 1. This suggests phenols to be a primary product within the reaction network. Multiple reaction pathways yielding the various products can of course occur in such a complex system, also with varying ranks, re- quiring further exploration [KHB12]. However, due to the complexity of this model and multitude of unknowns within the reaction pathways, use of the interpretation extensions could not be applied. Competing and parallel reactions must be expected, but the analysis performed here is restricted to identification and description of the major pathways at given reaction conditions, while still recognising that further reactions exist. Reviewing the plots of the primary product 4-ethylphenol suggests multiple pathways of quantitative significance that could potentially also produce other phenols directly from the biomass as primary products. The interpretation for this component is tentative, as the extrapolation based upon four data-points introduces a large degree of uncertainty in the determination of the y-axis intercept. The second rank Delplot in Figure 6.6b shows that all y-axis in- tersections are now finite, thus categorising catechols as secondary products.

Formulating a model pathway In terms of the wide product spectra observed in these reactions, a lump- model is expected to be the best approach and the analysis is focused on identifying and describing the significant reaction pathways. Single components which have been quantified in the course of the experimental work were used in the modelling, as well as global lump components. This allows the mapping of the complete pathway from the biomass to specific target compounds, such as phenol. Observations made in the experimental part of this study, as well as literature knowledge, are combined to formulate the model presented in Figure 6.7. Under given temperature conditions, lignin is seen to depolymerise effi- ciently and quickly. Solid residue detected after more than two hours residence time is assumed to mainly originate from repolymerisation reactions of phenolic compounds, such as catechols or phenols [McM+04]. Guaiacol has been stipulated as one of the key intermediates that forms phenols and catechols from lignin. The reaction pathway from methoxyphenols to catechols (k2) yields additional gas which is not further characterised since detailed information about the gas composition are not considered by the formal kinetic model. The amount of the gas produced from these reactions as well

54 Chapter 6. Kinetic studies of lignin degradation

as from the formation of phenol [JK85](k4) or char (k5) and dealkylation reactions (k8, k14) is however very small relative to the gas which originates from the gasification of formic acid and ethanol (k12, k13). Alkylation and dealkylation reactions (k7-k9, k14) are very dependent on the substituents attached to the aromatic ring [HKB12], and prior studies suggest that alkylation reactions would be largely preferred by guaiacols [Mil+99]. The abundance of 4-methyl and -ethyl substituents observed for guaiacols, which are also present in catechols and phenols, suggests the alkylation to take place either immediately from reactive species during the depolymerisation process, or as a very fast reaction from guaiacol, render- ing 4-ethylphenol to be the most abundant of this compound class. Either way, the initial residence time step chosen is not able to describe this reaction, and as the alkylation pattern of the para- position is subsequently seen in both catechols and phenols, lumping of the different alkylations is used. 2-Ethyl substitution is however, in significant quantities only found for phenols. Thus, a separate ortho- alkylation is proposed for 2-ethylphenol (k8), to explore the importance of this reaction. Furthermore, since dealkylation of 4-ethylphenol and alkylation of 2-ethylphenol is considered, the consideration of the reverse reactions, that are alkylation of 4-ethylphenol and dealkylation of 2-ethylphenol, is reasonable. All gaseous and inorganic liquid products, namely CO2, CO, H2,H2O and hydrocarbon gases (HC-gases), such as CH4,C2H4,C2H6,C3H6,C3H8, C4H10 were grouped under the model lump component «gas». Furthermore, LD (depolymerised lignin) is a lump component containing all components derived from lignin that are not detected or quantified byGC, but however, are soluble in ethanol, i.e. can not be categorised as solid residue or gas.

55 6.1. Lignin depolymerisation in ethanol in a batch-reactor

12 13 insoluable - k k

hydrocarbon lignin

O, O, 2 ethanol

, H 2 Gas

acid 13

EtOH 12 k

residue k , CO, CO , CO, Ethylphenol 2 - Phenols X Depolymerized H gases Solid Formic Ethanol

EtPhe - 11 D PhOH X L Gas k Char FA EtOH

D L

16 k = 633 K. 10 k

Char Gas 15

3 Gas k - Catechols

2 k

1 PhOH

k Lignin 17

Methoxy k 4 Gas k Gas

9 Gas EtPhe k 14 - Gas k 2

8 PhOH EtPhe

6 k - 7 k 4 k

5 k Figure 6.7: Scheme ofethanol main and reaction formic participants acid; and14 gaseous pathways and products of 17. of the This ethanol depolymerisation is and of indicated formic wheat by acid straw curved gasiﬁcation lignin arrows; take in part in reactions 2-4, 7-9,

56 Chapter 6. Kinetic studies of lignin degradation

6.1.4 Results from model fit The formal kinetic rate coefficients of the determined reactions were optimised using the algorithm explained in AppendixC. Different variations of the model shown in Figure 6.7 were considered, including additional reaction pathways, among them the direct conversion of methoxyphenols into phenols. After evaluating the results of the minimum search, the model version presented in Figure 6.7 was concluded to be the most appropriate. The focus was put on an acceptable mathematical fit for phenolic products. For the evaluation of the quality of the model fit the coefficient of determination R2 and the standard deviation σ were calculated as described in Appendix C.4. The results are shown in Table 6.2. The standard deviation of the experimental results determined by replicate experiments is depicted by the error bars in Figure 6.8 and 6.9. The coefficient of determination R2 is above 0.5, thus acceptable, for all phenolic products except 2-ethylphenol. The latter component shows a comparably high standard deviation and a low coefficient of determination. The model fit for the yield of gas and and solid residue are also acceptable since the coefficient of determination is high. Not only the development of the phenolics yield could be described (see Figure 6.8), the model also allowed to describe the yields of lumped gaseous products and solid residue (see Figure 6.9). Comparing the three rate coefficients which represent the three primary reactions of lignin degradation (k1, k6 and k10) it can be observed that the reaction to LD (k10) is the fastest. The fit curve for lignin shows that a total conversion of lignin can be assumed after 100 min to 120 min, which is in good agreement to the results obtained from the FT-IR analysis of the solid residue.

Table 6.2: Coeﬃcient of determination R2 for the model ﬁt and standard deviation σ of calculated and experimental yields (see Appendix C.4). Methoxy- PhOH Catechols 4-EtPhe 2-EtPhe Char Gas PhOH R2 0.96 0.59 0.88 0.57 -0.21 0.89 0.88 σ 14 44 20 30 113 56 26

The rate coefficients of the reaction of catechol to phenol (k4) is very low compared to other coefficients within the model and therefore of minor importance. The same is true for the gasification of phenols (k5). Ethylation of phenols is the only reaction pathway which is assumed to be reversible as described above. The rate coefficients of ethylation and deethylation are of the same order of magnitude. The ethyl substitution in the para- posi-

57 6.1. Lignin depolymerisation in ethanol in a batch-reactor

1 2 M e t h o x y P h O H C a t e c h o l s P h O H ) n i

n 2 - E t P h e g i l 8 4 - E t P h e g / g m (

d 4 l e i Y

0 0 2 0 0 4 0 0 1 0 0 0 1 2 0 0 R e s i d e n c e t i m e ( m i n )

Figure 6.8: The experimental values are compared with the model ﬁt functions for the quantiﬁed key phenolic compounds; T = 633 K.

1 . 0 S o l i d r e s i d u e 4 )

n C h a r i n

g L i g n i n ) i l n i g n

/ G a s g i l g ( g

/ e g ( u 0 . 5 2

d d i l s e i e r y

s d i l a o G S 0 . 0 0 0 2 0 0 4 0 0 1 0 0 0 1 2 0 0 R e s i d e n c e t i m e ( m i n )

Figure 6.9: Fits of the gas-lump and the solid residue compared with the experimental data. The diﬀerentiation between lignin and char in the solid residue was accomplished largely by FT-IR analysis, as illustrated in Fig- ure 6.4; T = 633 K.

58 Chapter 6. Kinetic studies of lignin degradation

tion (k14) is slightly faster than deethylation (k7), whereas the ethylation in ortho-position (k8) is about three times slower than deethylation (k9). Char formation from LD (k16) is of minor importance compared to the gasiﬁcation of LD (k11). Char formation from catechol (k3) on the other hand is faster than from phenol (k17) or from LD.

Table 6.3: Formal kinetic rate coeﬃcients (kj) of all 17 reactions deﬁned within the developed model (see Figure 6.7) kj k1 k2 k3 k4 k5 (min−1) 7.94·10−04 8.65·10−03 1.09·10−02 2.24·10−06 5.81·10−15

kj k6 k7 k8 k9 k10 (min−1) 3.61·10−04 1.79·10−02 1.92·10−03 5.71·10−03 4.68·10−02

kj k11 k12 k13 k14 k15 (min−1) 8.32·10−04 4.99·10−04 3.56·10+26 2.04·10−02 3.56·10−05

kj k16 k17 (min−1) 4.16·10−05 1.17·10−03

Within the model, polyaromatic components derived from the biomass (LD) are assumed to decompose directly into gaseous components (k11), but can also form methoxyphenols (k15) and char (k16). Mathematical fitting showed that the latter route is of minor importance. The fit also confirms that the degradation of formic acid is very fast at given temperatures (k13). The degradation of ethanol (k12) is significantly slower. However, at long residence times it is of importance, if a recovery of the solvent ethanol is considered for a technical process. This is in good agreement to the results obtained from the thermodynamic studies (see Chapter3).

6.1.5 Discussion Replicate experiments showed the averaged percentage error for the measured global lump products, that are solid residue and gas, to be 13 wt% and 10 wt% respectively. Standard deviations for these two average values were calculated to be 6 % and 2 % respectively. A possible explanation for this error is found in the determination of the solid residue moisture. Al- though the solids were weighed before and after drying, it is likely that a certain amount of the volatile moisture attached to the solid residue evaporated during separation of liquids and solids before weighing. Gas chromatographic analysis of the liquid product gives results for the yield of 12 diﬀerent kinds of phenolic components, certainly less than the

59 6.1. Lignin depolymerisation in ethanol in a batch-reactor real number of different phenolic products from lignin depolymerisation, but sufficient to evaluate a simple reaction network as suggested by Jegers and Klein for the pyrolysis of lignin [JK85]. The quality of results is supported by the acceptable fit of the single measurement data points in respect to the calculated yields as shown in Figure 6.10. The better the fit, the closer the data points plot to the diagonal line. The coefficients of determination R2 is above 0.5, thus acceptable, for all components except 2-ethylphenol. The latter component could hardly be detected in the liquid product. Thus the error analysis is very sensitive to slow changes of the yield or concentration which yields a high standard deviation (113 %) and a low coefficient of determination (-0.21). However, important for the validation of the model is also that the tendency of the development of yield over reaction time can be described as basically different from 4-ethylphenol.

1 . 0 2 - E t P h e 4 - E t P h e

s d l ) e i d y

e l 0 . 5 s i a l t a n e m r m i o

r M e t h o x y P h O H n e ( C a t e c h o l s p

x P h O H E 0 . 0 0 . 0 0 . 5 1 . 0 C a l c u l a t e d y i e l d s ( n o r m a l i s e d )

Figure 6.10: Comparison of calculated and experimental data for product yields of all phenolic (lump) components. Normalising factors (mg/glignin): MethoxyPhOH: 14.0, Catechols: 7.0, PhOH: 3.0, 4-EtPhe: 5.0, 2-EtPhe: 1.5.

Results from the FT-IR analysis of the solid residue as well as results from the ﬁt of the rate coeﬃcients show that approximately 100 min after starting the reaction, most of the lignin structure has disappeared. Solvolytic cleavage of the ether bonds present in the lignin structure, as already postulated for lignin degradation in a mixture solvent of ethanol and water by Yuan et al., is assumed to be the preferred reaction mechanism [Yua+10].

60 Chapter 6. Kinetic studies of lignin degradation

The observed effects of water on guaiacol pyrolysis describes a further parallel pathway based on the solution of the guaiacol molecules, thus pre- venting these from participating in free-radical reactions in order to form high molecular weight material [LK85]. This reaction pathway is believed to be transferable also to other protic solvent systems as well as lignin. Further- more, external hydrogen donor sources have been reported to cap the radical moieties with hydrogen, thus reducing the radical concentration, which can be advantageous and suppress repolymerisation reactions [Dor+99]. The reaction coefficient of the degradation of guaiacol (k2) under pyrolytic conditions at 633 K (circa. 0.0084 min−1) obtained by Vuori [Vuo86], is comparable to the value calculated within the fit of this model (0.0087 min−1). The good agreement of one of the calculated rate coefficients with those found in literature for similar reactions shows that the model is not only a mathematical method for the simulation of experimentally determined values, but also sensible in respect to the chemistry of lignin degradation. 4-Ethyl groups are generally the dominant substituents for all species at lower residence times. The occurrence of 4-ethylphenol as a primary product strongly suggests that ethyl group substituents are both a product of alkylation reactions in ortho- and para- position [HKB12] and depolymerisation. Supported by the Delplot analysis, it can be concluded that 4-ethylphenol is a primary product of lignin degradation. 2-Ethylphenol, however, is a late product and its yield hardly decreases toward very high residence times. It is thus assumed that 4-ethylphenol is formed directly from the lignin structure whereas 2-ethylphenol is formed by ethylation of phenols. Hydrogen donor solvents have been seen to influence various biomass degradation processes. The high yields of ethyl substituted guaiacols, and indeed also catechols and phenols at lower residence times, may be related to the capping of highly reactive vinyl- (to ethyl-) substituted guaiacol intermediates by these donor solvents [PK84; Ye+12]. This could be part of the reason for the high yield of 4-ethyl substituents for the different phenolic species, which in the work of Ye et al. comprised a total of 30 % (together with 4-vinyl substitutes) of all lignin derived components quantified [Ye+12]. The comparison of polymerisation pathways from catechol (k3) and phenol (k17) shows that catechols are more prone to repolymerise with k3 being greater than k17. This agrees with the results obtained by McMillen et al. [McM+04], who found that especially cresols are able to cap the radical induced coupling of hydroxybenzenes under pyrolysis conditions. The low reaction rates for gas formation from phenol (k5) furthermore supports the suggestion that mono-hydroxyl phenols are stable, contrary to the reactive methoxyphenols and catechols [JK85]. The model fits experimental values well and the derived rate constants

61 6.2. Lignin depolymerisation in ethanol in a continuous reactor match values found in literature. The number and choice of selected lumps, as well as their interaction pathways, are suﬃcient to describe the degradation of lignin with a focus on the deoxygenation and ethylation reactions in the production of mono-aromatics under the given reaction conditions. Alteration of temperature in the range of 643 K to 663 K [Kle+11] has been reported to show some systematic variations in the product spectrum. The model is based on to the use of ethanol as solvent medium as a source for ethylation reactions. The general degradation pathways shown in this section are however also found for alternative hydrogen donor systems [For+12]. The necessity of the introduced single 4-ethylphenol component, suggests that adaptations to the model for diﬀerent types of lignin would be necessary.

6.2 Lignin depolymerisation in ethanol in a continuous reactor

Based on a selected number of experiments in aMA1 and a continuous stirred tank reactor (CSTR) the applicability of the kinetic model developed in Section 6.1 to a different reactor system, that is a CSTR, and to varying reaction temperatures is explored. The applicability of the model to these alternate reaction conditions was tested and the fit quality was evaluated using sensitivity and flux analysis (see Appendix C.5 and C.6). Furthermore, the influence of temperature and the characteristics of a continuous reactor system on both a molecular as well as a global lump level was monitored. This aids in the understanding of the chemical reaction mechanisms and sets a basis for future investigative work.

6.2.1 Experimental Experiments using comparable loading ratios and identical reaction temperature were performed in both,MA1 (Exp. 56-61) and CSTR(Exp. 62-64, 68; see Appendix E.1). For different experiments in the CSTR the pressure and the mass flow m˙ was varied. These two parameters influence the mean residence time in the CSTR (see Table 6.4). Important for the further observations is the density inside the reactor at different reaction conditions. The density is proportional to the total mass of reactants inside the CSTR m.

m = ρ · VR,VR = 200 mL. (6.1) The density ρ is assumed to be constant throughout the experiment and

62 Chapter 6. Kinetic studies of lignin degradation is estimated employing the density of ethanol at the given pressure and temperature [Baz+07; DP04]. Since the reaction volume VR is equal to the volume of the reactor, the reaction mass m results to be constant throughout the experiment. Further experiments were conducted in the CSTR varying the temperature in the range between 633 and 673 K. The formic acid to lignin ratio of was 1.1:1 in the batch reactor and 1.2:1 in CSTR. The ethanol to lignin ratio was 9:1 for the batch-experiments and 9.5:1 in the experiments conducted in the CSTR.30

Table 6.4: Experimental conditions, estimated total reaction mass (m) and mass ﬂow (m˙ ) for batch and continuous reactor. Exp. Reactor T Pressure (mean) Resi- m m˙ dence time K MPa min g g/h 56/57MA1 653 - a 150 3.9 58/59MA1 653 - a 60 3.9 60/61MA1 653 - a 140 3.9 62 CSTR 653 25 52 62.8 72.8 63 CSTR 653 30 59 72.3 74.1 64 CSTR 653 20 39 50.2 75.6 65 CSTR 633 25 56 69.7 74.6 66 CSTR 673 25 44 56.2 76.3 67 CSTR 653 25 37 62.8 102.5 68 CSTR 653 25 21 62.8 180.4 a The design of theMA did not enable the measurement of resulting pressures. Comparable experiments have shown pressures between 29 and 33 MPa [Gas+10]. The pressure in presented experiments is expected to be reduced by 3-6 MPa due to a decreased FA/EtOH loading ratio.

6.2.2 Results General tendencies In this section experiments carried out in aMA1 (Exp. 58, 59) and a CSTR (Exp. 63) at 653 K and approximately 60 min (mean) residence time are compared and evaluated. The mass balance diagram in Figure 6.11a compares the mass fraction of the three diﬀerent output phases with the mass fractions of the input substances, that are ethanol, lignin and formic acid. In case of the batch-experiment (Exp. 58) inMA1 the results after 60 min residence time were considered. In case of the CSTR experiment, the exper-

63 6.2. Lignin depolymerisation in ethanol in a continuous reactor

1 0 0 1 0 0 7 0 a ) A b ) c ) F H C - G a s H 6 0 8 0 ) 8 0 2 )

% C O ) ( 5 0

n % C O i ( n

n 2

g i l o

n 6 0 6 0 i 4 0 t g o H / i c

t L o s s O g a t c r

E G a s 3 0 a m

4 0 F 4 0 r (

L i q u i d e

C a t e c h o l s d l s S o l i d 2 0 s e m M e t h o x y - i a 2 0 u 2 0 l P h O H Y 1 0 o M

V P h O H g i

0 L 0 0 L o a d i n g M A 1 C S T R M A 1 C S T R M A 1 C S T R

Figure 6.11: Results of lignin degradation inMA1 and CSTR at 653 K and 60 min (mean) residence times (Exp. 58, 63); a) liquid, gas and solid reaction products compared on a wt% basis with the input amounts of lignin from wheat straw Lig, EtOH and FA, b) Composition of the product gas, c) yields of PhOH, methoxyPhOH and catechols. imental results obtained in the stationary state (see Chapter 2.2.2) of the experiment with the corresponding mean residence time (59 min, Exp. 63) were observed. The diagram reveals that a clear reduction of the relative yields of gas and recovered solids can be achieved by transferring reactions from a batch to a continuous reactor setup. The mass balance shows a mass loss for the batch experiments of < 7 wt% and for the CSTR experiments of < 6 wt%.

The gaseous products comprise a mixture of H2, CO, CO2 and hydrocarbons in the C1 to C4 range, of which C2H6 and C2H4 are the largest contributors. A Comparison between batch and continuous operation shows both considerable impact on the quantity of gaseous products as well as on the gas composition (see Figure 6.11b). A considerable decrease of CO2 by a factor of 2, a 2 to 4 fold increased amount of CO and a large increase of H2 are observed in continuous operation. Whereas in batch processes, hydrogen is readily consumed [Gas+2], the surplus of H2 in the CSTR is a product of the continuous feeding, but also continuous tapping of gaseous components from the reactor. In addition, a control experiment was carried out in aMA1 at 653 K and 11 h residence time without biomass (Exp. 55). The input comprised 89 wt% ethanol and 11 wt% formic acid. The product contained 73 wt% liquid and 27 wt% gas and the main gaseous components were CO2,C2H6,C2H4,C4H10 that contribute with, respectively, 53 wt%,

64 Chapter 6. Kinetic studies of lignin degradation

13 wt%, 9 wt% and 13 wt% to the gaseous product. Also H2, CO, CH4 and C3-gases were detected. The high amount of C2-gases demonstrates that the gasiﬁcation of ethanol also contributes to the elevated gas production (see Chapter3. To be able to include not only global lump products, but to also cap- ture the role of typical deoxygenation and demethanisation reactions between monomeric phenolics in the liquid phase, 12 selected, 2- and 4-methyl, respectively -ethyl, substituted as well as non-substituted monomeric key components were quantitatively determined and grouped according to their structural properties, i.e. in accordance with their degree and type of oxygenation as described in Section 6.1. The yields for the grouped components per g initial lignin are shown in Figure 6.11c. The yield of each phenolic compound is given in Appendix E.1(Exp. 58, 63). The total yield percentage of all key compounds lies between 3 and 7 wt% on an initial lignin basis. As only a comparatively small number of key compounds were chosen, the obtained percentage of monomeric compounds in the product is an underestimation of the actual content. For all of the three grouped compound classes of phenolic compounds, that are methoxyphenols, catechols and phenols, a dominance of 4-ethyl substituents is observed. The total yield of phenolics can be increased by about 1 wt% from approximately 5 to more than 6 wt% on an initial lignin basis, if the reaction is transferred from a batch reactor to a CSTR. Especially the yield of phenols and methoxyphenols is increased, the yield of catechol is similar at about 60 min (mean) residence time.

Model fit and comparison The adapted kinetic model was used to fit the results of the temperature dependent experimental series (Exp. 64-66). The experimental results for the transition phase between non-stationary and stationary state in the CSTR are represented by the shattered data points in Figure 6.12. The continuous curves represent the results from the model fit. The modelling was carried out applying the methods presented in AppendixC. In order to minimise the the difference between calculated and experimental yields, the formal kinetic parameters EA,j and Aj were optimised (Equations C.1 to C.3). For the comparison of experimental and calculated yields of the phenolic products, see Figure 6.16. The coefficient of determination R2 and the standard deviation σ of the experimental yields from the calculated yields of the phenolic products is listed in Table 6.5. The model fit of the experimental results of 2-ethylphenol was difficult which is discussed in Section 6.1. Be- sides, also the model fit of 4-ethylphenol and catechols was critical as the low coefficient of determination shows. The resulting formal kinetic rate co-

65 6.2. Lignin depolymerisation in ethanol in a continuous reactor a ) 5 0 b ) 5 0

) 4 0 4 0 n i n g i l

g 3 0 3 0 / g

m 2 0 2 0 (

d l 1 0 1 0 e i Y 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 c ) 5 0 E x p e r i m e n t a l d u r a t i o n ( m i n )

) 4 0 n i M e t h o x y P h O H n g i l C a t e c h o l s g 3 0 /

g P h O H

m 2 0 ( 4 - E t P h e

d l 2 - E t P h e

e 1 0 i Y 0 0 1 0 0 2 0 0 3 0 0 4 0 0 E x p e r i m e n t a l d u r a t i o n ( m i n )

Figure 6.12: Comparison between the CSTR-ﬁt of phenolic lumps and their measured quantities for Exp. 64-66 conducted at a) 633 K, b) 653 K and c) 673 K; mean residence times are between 39 and 56 min (see Table 6.4).

efficients kj, apparent activation energies EA,j and the Arrhenius factors Aj are listed in Table 6.6. The fits for both CSTR and batch experiments from Section 6.1 were evaluated using flux and sensitivity analysis (see Figure 6.13 and Figure 6.14). Comparisons between the single formal kinetic rate coefficients are given in the last column in Table 6.6. If the quotient of kj(CS,633 K)/kj(MA,633 K) is smaller than 0.02 or greater than 50, the field is left blank since such values simply show a great difference between the CSTR-fit and theMA-fit. Nevertheless, this difference is discussed in Section 6.2.3 and an eventual disagreement is pointed out. These show that the resulting formal kinetic rate coefficients are similar for both fits in respect to most reactions involved in the formation and degradation of phenolics (k1, k2, k5, k6, k10). The reaction rates of all these reactions increased by transferring them to a CSTR, however, the maximum increase factor was 3.3. Among the reaction rates of

66 Chapter 6. Kinetic studies of lignin degradation

Table 6.5: Coeﬃcient of determination R2 for the model ﬁt and standard deviation σ of calculated and experimental yields of phenolic products (see Figure 6.12). methoxyPhOH PhOH catechols 4-EtPhe 2-EtPhe R2 0.80 0.62 0.55 0.31 -0.55 σ 18 41 37 30 118

the primary reactions (k1, k6 and k10) the increase factors were even smaller, 3.1 for the formation of methoxyphenols from lignin (k1), 1.3 for the formation of 4-ethylphenols (k6). Also the fluxes of these reactions were similar for both fits. However, the fluxes for all three reactions (r1, r6 and r10) were smaller for the CSTR fit (see Figure 6.13), the flux of r6 was reduced by a factor of 2.2 from 8.0·10−4 to 3.6·10−4 g/mL. On the contrary the formation of methoxyphenols from LD as well as the conversion of catechols into phenols, respectively r15 and r4, were more than 30 times faster in the CSTR according to the model fit. The fluxes of both reactions increased of −3 −3 a factor greater than ten to 3.0·10 g/mL and 1.1·10 g/mL for r15 and r4 respectively. Hence, the sum of the fluxes which lead to a formation of methoxyphenols (r15 and r1) at 60 min (mean) residence time more than doubles if the reaction is transferred from a batch to a continuous reactor. In addition, the flux of the conversion of methoxyphenols into catechols (r2) also doubles. The flux from lignin to phenol via methoxyphenols and catechols is thus favoured by a continuous reactor according to the results from modelling. The reversible reactions are obviously in equilibrium since the fluxes of these reactions (r7 and r14 as well as r8 and r9) are equal (see Fig- ure 6.13). The formal kinetic rate coefficients of the char formation from catechols (k3), phenols (k17) and LD (k16) significantly decrease by a factor of more than 10 when the reaction is transferred to a CSTR. Contemplat- ing the flux of the repolymerisation of phenols (r16), which is relatively low (< 10−4 g/mL), it is easily visible that this reactions plays a minor role in both reactor systems. On the contrary the fluxes of the char formation from LD (r17) and the repolymerisation of catechols (r3) are both reduced from respectively 1.9·10−4 to 1·10−4 g/mL and 1·10−4 to < 10−4 g/mL.

67 6.2. Lignin depolymerisation in ethanol in a continuous reactor j b A (CS,633 K)/ (MA,633 K) j j k k 3.1 1.8 35 3.3 1.3 2.0 0.3 30 0.1 0.1 b 4 3 4 3 4 2 3 2 2 4 4 2 5 4 4 − − +0 − − − − − − − − − +8 − − − − and Arrhenius factors 1 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 · · · · · · · · · · · · · · · · · − A,j (MA,633 K) j E min k 7.95 8.24 2.02 7.97 2.47 3.75 3.27 8.74 2.71 4.71 1.62 2.69 1.09 4.20 3.04 4.18 5.41 b 3 2 2 3 4 2 5 4 5 5 − − − − − +0 +2 +3 − − +0 +0 − − − 6 6 1 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 · · − · · · · · · · − · · · · · · − (CS,633 K) 10 10 j k min 2.48 1.51 < 2.80 8.25 4.87 2.92 4.30 4.29 9.43 < 7.70 1.55 7.22 9.22 3.20 3.22 b 3 2 2 2 4 1 4 3 5 5 − − − − − +0 +2 +3 − − +0 +0 − − − 6 6 1 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 · · − · · · · · · · − · · · · · · − (CS,653 K) 10 10 j , apparent activation energies k 4.34 min 2.51 < 2.94 1.01 6.77 3.16 4.32 4.29 1.46 < 2.30 1.55 7.22 1.02 6.76 6.65 j k b 3 2 2 2 4 1 4 3 4 4 − − − − − +0 +2 +3 − − +0 +0 − − − 6 6 1 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 · · − · · · · · · · − · · · · · · − (CS,673 K) 10 10 j k min a ) 1 0 − log(A/A . 1 − A E 96 5.30 7.32 kJ/mol min 87 5.40 4.06 (528) 1.61 < 8 -0.88 3.07 34 0.72 1.21 57 1.35 9.22 14 1.60 3.40 1 2.72 4.35 (0) 3.63 4.29 75 5.18 2.20 (0)188 -9.28 11.40 < 6.43 (0) 0.21 1.56 (0) 0.86 7.22 17 -1.67 1.11 129 6.12 1.37 125 5.79 1.31 =1 min 0 A CS : continuousMA reactor, : microbatch-autoclave. j 1 2 3 4 5 6 7 8 9 r r r r r r r r r 10 11 12 13 14 15 16 17 r a b r r r r r r r r Table 6.6: Formal kineticof rate coefficients the 17 reactionscompared defined with rate in coefficients the calculated model for (see the Figure batch-fit6.7 ) at at 633 K. different temperatures for the CSTR -fit are

68 Chapter 6. Kinetic studies of lignin degradation

Lignin  MethoxyPhOH Batch MethoxyPhOH  Catechols CSTR Catechols  Char Catechols  PhOH

PhOH  Gas Lignin  4-EtPhe 4-EtPhe  PhOH PhOH  2-EtPhe 4-EtPhe  PhOH

Lignin  L D LD  Gas EtOH Gas FA  Gas PhOH  4-EtPhe

LD  MethoxyPhOH LD  Char PhOH  Char 10-4 10-3 10-2 10-1 100 101 Flux (g/mL)

Figure 6.13: Comparison of the fluxes of each reaction defined in the model (see Figure 6.7) calculated for both batch-fit and CSTR-fit at 60 min (mean) residence time τ (τ 0) and 633 K (see Appendix C.6).

The formal kinetic rate coefficient of the gasification of LD (k11) decreases by a factor greater than 50, its flux decreases from a relatively high value in a batch reactor (3.2·10−3 g/mL) to a negligible value (< 10−4 g/mL) in a continuous reactor. The flux of the gasification of ethanol decreases from 0.022 g/mL to 3.0·10−3 g/mL if the reaction is transferred fromMA to CSTR. The formal rate coefficient for the gasification of phenols is increased by a factor of 3.3. However, the flux in a CSTR of this reaction at 60 min residence time is still negligible. The flux of the gasification of formic acid is similar for both, the CSTR-fit and the batch-fit, although the formal kinetic rate coefficient (k13) calculated by the CSTR-fit and the batch-fit differ from each other by a factor more than 50. The gasification of formic acid (r13) is, if the reversible reactions are not considered, the fastest reaction in both fits. The sensitivity analysis as explained in Appendix C.5 yields a matrix showing the sensitivity of each defined component to all considered reactions (see Figure 6.14). If none of the components is sensitive to a defined reaction, the formal kinetic rate coefficient can not be optimised mathematically since a modification of the formal kinetic rate coefficient does not show an effect on the yield of any component. Nevertheless the reaction might be essential

69 6.2. Lignin depolymerisation in ethanol in a continuous reactor for the model since it is the only pathway leading to a certain component. Figure 6.14 shows that all phenolics are sensitive to the primary formation of methoxyphenol (r1) and its most important competing reaction (r10). However, the sensitivity to r10 of all phenolic products is reduced in the CSTR-fit compared to the sensitivity in the batch-fit. Major differences between the batch and the CSTR fit were found in respect to the char yield, which is only sensitive to the formation of char from LD (r16) in the CSTR-fit and to the char formation from catechol (r3) as well as to the preceding reactions (r1, r2) in the batch-fit. Noteworthy is the increased sensitivity of all phenolics to the formation of methoxyphenols from LD, which is not existent in the batch fit, however, significant in the CSTR fit. A very similar observation can be made for the sensitivity of catechols, phenols and ethylated phenols to the conversion of catechols into phenols. On the contrary the sensitivity of phenols and ethylated phenols to the formation of 4-ethylphenol from lignin is higher in the batch-fit compared to the CSTR-fit.

Temperature dependence of formal kinetic parameters When increasing the temperature in the CSTR from 633 to 673 K, decreasing amounts of solids (-1.5 wt% on a global mass scale) and increasing amounts of produced gas (+9.9 wt% on a global mass scale) were recovered (see Appendix E.1). The yields of all phenolic products increased with the temperature except the yield of methoxyphenols, which were found to decrease. This tendency is correctly described by the model ﬁt (see Figure 6.12). Fig- ure 6.15a shows that the initial slope of the curve describing the yield of methoxyphenols is similar for all temperatures. It further reveals that the yield of methoxyphenols is reduced by about 1/3 for 673 K in respect to the maximum yields achieved at 633 K. The yield of catechols is overestimated by the model for 633 K and underestimated for 653 K (see Figure 6.15b). The slope describing the increase of the yield of catechols and phenols during the non-stationary state increases with temperature. Higher yields of both compound classes are found for higher reaction temperatures. Noteworthy is the almost threefold increase of the catechols yield within the contemplated temperature range (633 to 673 K). In addition the catechols yield at 633 K seems to have not reached the stationary state until 360 min experimental duration.

70 Chapter 6. Kinetic studies of lignin degradation Figure 6.14: Sensitivity60 on min a (mean) scale residence time from to 0 a to variation 100 of % rate coefficients (see by Appendix factorC.5 ) 2 of for a) the batch-fit calculated and yields b) atCSTR -fit. 633 K and

71 6.2. Lignin depolymerisation in ethanol in a continuous reactor

) a ) b ) n i 4 0 n 2 0 ) g i l n i n g / g i l

g 3 0

g 1 5 / m g (

m H 2 0 (

1 0 O s l h o

P 6 7 3 K h

y 1 0 5 c

x 6 5 3 K e o t h 6 3 3 K a t 0 C

e 0

M 0 1 0 0 2 0 0 3 0 0 4 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 E x p e r i m e n t a l d u r a t i o n ( m i n ) E x p e r i m e n t a l d u r a t i o n ( m i n )

Figure 6.15: Results from the CSTR-ﬁt of the experimental yields at diﬀerent temperatures and mean residence times (Exp. 64-66) of a) methoxyphenols and b) catechols.

The calculated apparent activation energies listed in Table 6.6 indicate the temperature dependence of each reaction. Alkylation and dealkylation reactions of phenols (reactions r7-r9 and r14) as well as the gasification of formic acid and LD show an apparent activation energy close or equal to zero, i.e. a negligible temperature dependence of the formal kinetic rate coefficient. However, these results have no influence on the description of the product yields since these reactions are reversible and in equilibrium as shown above. The reaction with the highest activation energy (r3) is of minor importance since the formal rate coefficients and fluxes are negligible. Comparing the apparent activation energies of the gasification of ethanol (r12) and formic acid (r13), it can clearly be observed that formic acid is instantaneously gasified at all temperatures, whereas the gasification of ethanol is much more sensitive to temperature changes. Hence, the main reason for the increased gas yield found at higher temperatures next to further cracking reactions [Kle+11], is due to the increased gasification of ethanol. The formation of methoxyphenols (EA1=96 kJ/mol) shows a higher apparent activation energy than the reaction of its decomposition (EA2=87 kJ/mol). The same phenomenon is observed in respect to catechols. The calculated apparent activation energy of the formation of catechols (EA2) is higher than for the conversion of catechol into phenol (EA4=8 kJ/mol) which is the major reaction of catechol decomposition.

72 Chapter 6. Kinetic studies of lignin degradation

6.2.3 Discussion

The CSTR-fit as well as the batch-fit of the formal kinetic rate coefficients show that the developed model shown in Figure 6.7 considers the main reaction pathways of wheat straw lignin degradation in ethanol with formic acid. The optimised formal kinetic parameters approximate the yields of the phenolic products, thus the model is able to describe the general tendencies of the phenolics yield in a CSTR (compare also Figure 6.16). Sensitivity and flux analysis demonstrate the applicability of the model for both reactor systems, batch reactor and CSTR. Comparison of the results from this work with literature values shows a good general agreement, and thus further confirm the validity of the developed model. Jegers and Klein found slightly varying estimated rate constants for catechol and guaiacol decomposition during Kraft lignin pyrolysis at 673 K, depending on their alkyl substitution. 4-Ethyl substituted catechols and guaiacols were seen to decompose faster than their 4-methyl substituted analogues, which in turn decomposed faster than the non-substituted compounds. Averaged, at 673 K rate constants of 0.029 min−1 (0.031 min−1 in this work) for catechols and 0.042 min−1 (0.041 min−1 in this work) for guaiacols were found. Pure guaiacol displayed a decomposition rate constant of 0.038 min−1 [KV08] at 673 K, illustrating the non-influence of other species on the decomposition rate. Various activation energies for lignin degradation have been reported by Brebu and Vasile [BV10]. The most common values for single first order decomposition are found to vary between 54 and 79 kJ/mol between 517 - 582 K, increasing to 81 kJ/mol at higher temperatures up to 1440 K [Ram70; DRS74; Nun+85]. The apparent activation energy describing the primary lignin degradation in this work, that is principally EA,10 (= 75 kJ/mol), is within this typical span. The model correctly describes the increase of yields of catechols and phenols and the decrease of the yield of methoxyphenols at increasing temperatures (see Figure 6.12). However, one phenomenon is observed that was not correctly described by the model. The experimental results at 673 K show that approximately 100 min into the experiment, the yield increase of catechols stagnates (see Figure 6.15 b) and the actual final yield at long experimental durations is seen to be very similar to the yield at 653 K. The model underestimates the catechol yield in the stationary state at 653 K since the slope of the catechol yield during the first 100 min of the experiment at 653 K is lower than at 673 K. In addition, the phenomenon leads to and overestimation of the rate coefficient of the formation of catechol (k2) at 633 K. The abrupt stagnation of the catechol yield is difficult to describe

73 6.2. Lignin depolymerisation in ethanol in a continuous reactor

M e t h o x y P h O H

) 1 . 0

d C a t e c h o l s e

s P h O H i l

a 2 - E t P h e

m 4 - E t P h e r o n (

l 0 . 5 e i y

l a t n e m i r e

p 0 . 0 x E 0 . 0 0 . 5 1 . 0 C a l c u l a t e d y i e l d ( n o r m a l i s e d )

Figure 6.16: Comparison of the normalised calculated and experimental data for product yields of all phenolic (lump) components. The following normalising factors (mg/glignin) were used for better comparability: MethoxyPhOH: 42.1; Catechols: 20.4; PhOH: 5.6; 4-EtPhe: 19.4; 2-EtPhe: 3.1.

by any formal kinetic model. Neither the consideration of additional intermediates, reactants or reaction pathways which are not considered by the model, nor increased reaction orders were able to improve the fit. A change of the reaction mechanism caused by e.g. phase separation in the reactor might be a reason. The effect of the density on the reaction mechanism might also play an important role. In this work, however, only the effect of the density on the residence time was considered. This leads to an underestimation of the yield o 4-ethylphenol at 653 K and thus a rather low coefficient of determination at this temperature while the approximation of the experimental yield of 4-ethylphenol by the model is acceptable at 633 K and 673 K. Further research effort has to be spent on the clarification of these phenomena. The study of ethanolysis and hydrothermal degradation of intermediate phenolic products, that are methoxyphenols, catechols and ethylphenol, are suggested.

Both model fits show that the main fluxes of the reactions r10, r12, r13 are in the same order of magnitude, whereas the gasification of ethanol seems

74 Chapter 6. Kinetic studies of lignin degradation to be favoured in the batch reactor. Small differences can be due to slightly different input concentrations and due to the assumption that all reaction orders are 1. The suppression of the gasification and the formation of char from LD can be explained by a parallel reaction from the reactive component LD to a stable, ethanol-soluble component LD. The latter obviously occurs within lignin depolymerisation but is not crucial for describing the experimental results. A CSTR facilitates reactions of early intermediates and later products, e.g. LD and phenols respectively. This might result in the scavenging of radicals and the suppression of char formation as described in other works [Sai+03; McM+04; Tak+12; Aid+02; Fan+05]. The sensitivity analysis shows a high sensitivity of all phenolic components to the formation of LD (r10), which is the reaction competing with the primary phenolic products methoxyphenols and 4-ethylphenol. Hence, the key for increasing the phenolic yield is to accelerate the formation of methoxyphenols (r1) and 4-ethylphenol (r6) and to suppress the competing parallel reaction (r10). However, the CSTR-fit revealed that another important reaction pathway to yield phenolics is via the reactive intermediates LD (r15) and that this pathway is favoured in a CSTR. The modelling reveals that in a CSTR the major reaction pathway from lignin to phenol is via methoxyphenols and catechol, whereas in a batch reactor it is via 4-ethylphenol. The formal kinetic rate coefficients of the primary reaction of lignin depolymerisation (r1, r6 and r10) calculated for the CSTR-fit are higher than for the batch-fit (factor 1.3 to 3.1). These are the result of the impact of stirring and the increased heating rate in the CSTR. The temperature dependence of the formation of methoxyphenols and catechols (r1 and r2, respectively) is higher than those of the subsequent decomposition reactions (r2 and r4, respectively). This implies that increased temperatures lead to an increased yield of catechols and methoxyphenols. However, a temperature increase within the range of 633 - 673 K is not seen to aid in an accumulation of these monomeric units (see Figure 6.15a). This can be explained by a drop of the mean residence time with increasing temperatures due to changing densities. The calculated apparent activation energies of the alkylation and dealkylation of phenols (EA7-EA9 and EA14) are close to zero and thus the formal kinetic rate coefficients of these reactions are not sensitive to temperature changes. Obviously, the only important fact for modelling these reversible reactions in a CSTR is the thermodynamic equilibrium. Fluxes and the temperature dependence of the mentioned reactions are of minor importance. Furthermore, the gasification of formic acid (r13) and LD (r11) resulted to be insensitive to temperature changes (EA,j close to 0) whereas the tem-

75 6.3. Lignin depolymerisation in water in a batch-reactor

perature sensitivity of the repolymerisation of catechol (EA3) is very high (528 kJ/mol). These values are of minor importance due to an exceptionally high rate coefficient as for the gasification of formic acid (k13) or an exceptionally low rate coefficient as for the repolymerisation of catechol (k3). The sensitivity analysis shows that none of the reaction participants is sensitive to a change of k3 or k11. The gasification of ethanol is mainly influenced by residence time and temperature. The flux of this reaction is one of the highest at 60 min residence time. High temperatures and long residence times favour the gasification of ethanol. In the CSTR the flux of ethanol gasification is considerably lower (0.003 g/mL) than in a batch reactor (0.022 g/mL). Thus, the model fit implies a transfer from batch to CSTR to suppress the gasification of ethanol. The transfer from batch to CSTR showed a suppression of gasification and repolymerisation of reactive intermediates as well as of phenolic products. In addition the gasification of ethanol was reduced. The formation of methoxyphenols from reactive intermediates (LD) was revealed to be an important reaction pathway, favoured in a CSTR to increase the yield of phenolics.

6.3 Lignin depolymerisation in water in a batch- reactor

Hydrothermal degradation of lignin has become more important in the last years. Some research has been done, especially focusing on the reaction kinetics of the degradation of model compounds [WSG08; WSG09; WSG11]. Zhang, Huang, and Ramaswamy described the lignin degradation in hot compressed water and the build-up of char [ZHR11]. Yong and Matsumura [YM12] studied the reaction kinetics of lignin depolymerisation in SCW in a continuous reactor considering both, global lump components and single phenolic components. In this chapter the question should be answered whether the model shown in Section 6.1 can be adapted in order to describe the hydrothermal depolymerisation of enzymatic hydrolysis lignin from spruce wood (softwood) in a batch-reactor. Experiments (Exp 69-102, see Ap- pendix E.1) were carried out in a 5 mLMA2. The reactants for all experiments were 0.33 g of spruce-lignin and 2.5 g of water. Neither a hydrogen donor nor any homogeneous or heterogeneous catalysts were added. The modelling of the major reaction pathways and the formal kinetics were validated by comparing calculated and experimental yields of phenolics as well as solid and gaseous lump products. In addition, the formal kinetic parameters resulting from an optimisation were compared with values from literature.

76 Chapter 6. Kinetic studies of lignin degradation

- 13

k , O 2 via

partially H

interme

, residue soluble, not insoluble - - CO FID - O O

2 2 2 14 Phenols Reactive intermediates Stable diates H detectable GC Solid H k

CO, H CO,

D1 D2 PhOH L L Char

15 k

11 k

9 k

8 D2 k L

12 D1

k L 7 k

k O 4 Char 2 ,H

2 4 H , CH , 2

CH 2

5 H

Catechols CO k 2 k Lignin

2 k CO

PhOH CO -

1 k

,CH

2 O 2 CO ,H Methoxy 2

3 PhOH 6 k k Figure 6.17: Scheme oflignin; main reaction gaseous participants components and take pathways part of the in hydrothermal reactions depolymerisation 2-6 of and spruce- 13. This is indicated by curved arrows.

77 6.3. Lignin depolymerisation in water in a batch-reactor

6.3.1 Model adaptations

The development of the formal kinetic model describing the hydrothermal lignin depolymerisation in a batch reactor is based on the scheme of reaction pathways developed in Section 6.1. However, the model had to be adjusted to the application for the different solvent, i.e. water instead of ethanol, and type of lignin. The lignin used here was an enzymatic hydrolysis lignin from spruce (softwood) which is principally made up of coniferyl moieties. It was consequently observed that the most prominent initial products from spruce- lignin were methoxyphenols, while only small amounts of 4-ethylphenol were detected. Lignin from wheat straw, which was used in Section 6.1, on the contrary yields significant amounts of 4-ethylphenol since it contains more para-coumaryl units. Hence, for the modelling of the degradation of spruce- lignin a primary formation of 4-ethylphenol did not have to be considered. In addition, hydrothermal degradation of lignin yields high amounts of catechols compared to thermal degradation in ethanol or acetone [LZ08]. It is however impossible to describe such high yields by assuming the formation of catechols exclusively via methoxyphenols. A parallel pathway for the formation of catechol was considered. However, the primary intermediate of this pathway is unknown. It is supposed to be among the lump component named LD. Yong and Matsumura did not report increased catechol yields, thus did not consider this parallel pathway of catechol formation [YM12]. This is probably due to the low loadings of lignin (0.1 wt%). The lump component LD comprises all water soluble components that could not be quantified byGC-analysis as well as volatile organic components which were adsorbed on the solid residue and thus did not pass the filter during product separation but were evaporated during the drying of the solid residues. The component LD could analytically not be quantified, however, its existence is reasonable to be assumed in order to close the mass balance. Previous studies concerning repolymerisation reactions and scavenging of organic radicals [Far+10; Sai+03] led to the assumption that among the lump component LD reactive components (LD1) and stable components (LD2) can be distinguished. In addition, the gaseous products were further distinguished into reactive intermediates (CO, H2) and stable products (CO2 and CH4). Gaseous products were assumed to be formed via the gasification of organic components derived from lignin. Furthermore, methanisation and water-gas-shift- reaction were taken into account. These adaptions of the model shown in Figure 6.7 lead to a modified model which is assumed to be able to describe the reactions taking place throughout the hydrothermal depolymerisation of a softwood-lignin. The modified model is shown in Figure 6.17. The mod-

78 Chapter 6. Kinetic studies of lignin degradation elling of the kinetics of hydrothermal lignin degradation was carried out as described in AppendixC. All reactions were assumed to be of first order. The resulting differential equations are listed in Appendix C.3. Since the formal rate coefficients depend on the temperature, Equations C.1 to C.3 were applied for their optimisation.

6.3.2 Results The overall experimental results are summarised in Appendix E.1(Exp. 69- 102). The experimental yields of stable gaseous products (CO2 and CH4) and solid residue are displayed in Figure 6.18, represented by the individual data points. Results of hydrothermal lignin depolymerisation show a recovery of stable gaseous products of maximum 25 wt% on an initial lignin basis, of which 24 wt% are CO2. After an initial rapid increase of the yield of CO2 and CH4 until approximately 60 min residence time, the yield approximates a finite threshold value. This threshold increases with the temperature. The maximum yield of 15 mg/glignin CO was measured at 653 K and 30 min residence time (Exp. 95). It was observed to peak at residence times between 30 and 120 min. With increasing temperature the maximum of the peak increased and the peak shifted to shorter residence times (see Appendix E.1, Exp. 69-102). H2 could not be detected in the gaseous product, however, was assumed to be formed and instantaneously consumed, since several reactions which occur within hydrothermal lignin degradation require the presence of reactive hydrogen, e.g. the formation of catechol by demethylation of methoxyphenol [Vuo86] or the hydrolytic scission of one hydroxyl-group of the catechol molecule [WSG09]. After an initial decline until approximately 60 min residence times the yield of solid residue approximates a finite threshold value. This threshold varies marginally with the temperature. Not converted lignin and char could gravimetrically not be distinguished. The measured yield of solid residue was thus compared with the sum of the calculated yields of not converted lignin and char. The interpretation of the FT-IR-spectrum of the solid residue aids in analytically determining the zone of transition from not-reacted lignin to char. Figure 4.2b shows the FT-IR-spectra of spruce-lignin and the solid residue after different residence time in theMA1 at 633 K. Most aromatic vibrational bands, e.g. at 1511 cm−1 1429 cm−1 and 1058 cm−1, disappear until residence times of 60 min. The interpretation of the FT-IR-spectra is analogous to the lignin degradation in ethanol in Section 6.1. The yield of solid residue is described by the model assuming a primary decomposition of the lignin followed by the build-up of char. The standard deviation σ of experimental and calculated yields is 18 %. Gaseous products

79 6.3. Lignin depolymerisation in water in a batch-reactor

1 . 0 1 . 0 L i g n i n C h a r a ) b ) L D 1 L D 2 0 . 8 0 . 8 S o l i d R e s i d u e ) n i C O & C H n 2 4

g 0 . 6 0 . 6 i l g / g

( 0 . 4 0 . 4

d l e i 0 . 2 0 . 2 Y

0 . 0 0 . 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 1 . 0 c ) 1 . 0 d )

0 . 8 0 . 8 ) n i n

g 0 . 6 0 . 6 i l g / g

( 0 . 4 0 . 4

d l e i 0 . 2 0 . 2 Y

0 . 0 0 . 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 R e s i d e n c e t i m e ( m i n ) R e s i d e n c e t i m e ( m i n )

Figure 6.18: Comparison between the model-ﬁt of solid and gaseous lump components and measured quantities of solid residue and stable gaseous compounds (CO2 and CH4) for experiments carried out at a) 593 K, b) 613 K, c) 633 K, d) 653 K.

are assumed to be formed via the gasiﬁcation of LD1 and phenols. In addition, CO2 and CH4 were assumed to be produced during the decomposition of methoxyphenols and catechols (see Appendix C.3). The water-gas-shift reaction contributes to the conversion of CO into CO2 due to the excessive amount of water present in the reactor. The yield of the stable gaseous compounds CO2 and CH4 is underestimated by the model during the initial increase up to approximately 90 min. When the yield approaches the threshold, it is overestimated at any contemplated temperature. The standard deviation σ of experimental and calculated yields for stable and reactive gaseous compounds is 19 %. The total yield of phenolic products reached about 3.5 wt% of the initial input mass of lignin (see Figure 6.19). The dominant phenolic products were catechols, that are catechol, 4-methylcatechol and 4-ethylcatechol. The

80 Chapter 6. Kinetic studies of lignin degradation

0 . 0 3 M e t h o x y P h O H a ) 0 . 0 3 b ) C a t e c h o l s ) n i

n P h O H g i l 0 . 0 2 0 . 0 2 g / g (

d l

e 0 . 0 1 0 . 0 1 i Y

0 . 0 0 0 . 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0

0 . 0 3 c ) 0 . 0 3 d ) ) n i n g i l 0 . 0 2 0 . 0 2 g / g (

d l

e 0 . 0 1 0 . 0 1 i Y

0 . 0 0 0 . 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 R e s i d e n c e t i m e ( m i n ) R e s i d e n c e t i m e ( m i n )

Figure 6.19: Comparison between the model-ﬁt of phenolic lumps and their measured quantities for experiments conducted at a) 593 K, b) 613 K, c) 633 K, d) 653 K; error bars show standard deviation σ obtained from similar reproduced experiments at the corresponding residence time and temperature. yield of catechols increased with the residence time and the temperature approximated a threshold maximum yield. After an initial rapid increase of the yield of methoxyphenols it decreased with the residence time. The peak of the yield of methoxyphenols shifted from 90 min at 613 K to 30 min at 653 K. The higher the temperature the slimmer is the form of the peak. The yield of phenols increased with the residence time and the temperatures. In addition to the experimental yields (individual data points) the calculated yields are displayed in Figure 6.19 (continuous curves). The formation and degradation of methoxyphenols as well as the high yield of catechol can be described by the model. Furthermore, the increasing yield of phenols can be described by the model. The standard deviation σ of the experimental from the calculated yields of methoxyphenols, catechols and phenols are 12 %, 8 % and 15 % respectively, the R2 for these phenolics is 0.78, 0.88, 0.10

81 6.3. Lignin depolymerisation in water in a batch-reactor respectively. While R2 for methoxyphenols and catechols is acceptable, the approximation of the yield of phenols by the model is much worse especially for low yields of phenols. For relatively high yields of phenols, which were especially obtained at increased temperatures, a better approximation of the model fit could be achieved. For the yields of phenols from the experiments at 653 K R2 = 0.83. Table 6.7 shows all optimised formal kinetic parameters resulting from the search of the minimum difference between the calculated and the measured yields (see Equation C.1). The optimisation considered the formal kinetic rate coefficients of the formation of char (k5) and LD2 (k12) from catechols to be negligible. For all other reaction pathways sensible values of EA,j and Aj were calculated. The dominating primary pathway of lignin degradation is the formation of LD1 since the resulting formal rate coefficient k7 is about 100 times larger than the formal rate coefficient of the formation of methoxyphenols k1 for all considered temperatures.

Table 6.7: Apparent activation energies EA,j, Arrhenius factors Aj and rate coefficients kj at different temperatures for the 15 reactions defined in the model (see Figure 6.17).

0 a rj EA log(A/A ) k(593 K) k(613 K) k(633 K) k(653 K) kJ/mol - min−1 min−1 min−1 min−1 −4 −4 −4 −4 r1 58 1.4 1.91·10 2.81·10 4.02·10 5.63·10 −3 −3 −3 −2 r2 101 6.4 2.64·10 5.16·10 9.68·10 1.75·10 −4 −4 −3 −3 r3 94 5.0 5.15·10 9.61·10 1.72·10 2.98·10 −4 −4 −4 −4 r4 55 1.0 1.34·10 1.93·10 2.72·10 3.76·10 −6 −6 −6 −6 r5 3037 3.2 < 10 < 10 < 10 < 10 −5 −5 −5 −5 r6 66 1.0 1.84·10 2.85·10 4.27·10 6.26·10 −2 −2 −2 −2 r7 32 1.2 2.38·10 2.94·10 3.58·10 4.31·10 −3 −3 −3 −2 r8 81 4.6 3.52·10 6.01·10 9.90·10 1.58·10 −3 −3 −3 −3 r9 58 2.6 2.89·10 4.25·10 6.09·10 8.53·10 −4 −4 −4 −4 r10 72 2.5 1.55·10 2.49·10 3.89·10 5.90·10 −3 −3 −3 −3 r11 75 3.8 1.45·10 2.39·10 3.82·10 5.92·10 −6 −6 −6 −6 r12 296 5.0 < 10 < 10 < 10 < 10 −3 −3 −3 −3 r13 48 1.6 1.92·10 2.65·10 3.58·10 4.75·10 −2 −2 −2 −2 r14 42 2.2 2.71·10 3.59·10 4.68·10 5.99·10 −4 −4 −3 −3 r15 69 2.8 4.75·10 7.52·10 1.16·10 1.73·10 a A0 = 1min−1

82 Chapter 6. Kinetic studies of lignin degradation

6.3.3 Discussion

Comparison of ethanol and water model

For the comparison of the two different models, developed for the degradation of wheat-straw-lignin in ethanol and formic acid on the one hand and for the degradation of spruce-lignin on the other, a flux analysis for 633 K and 131 min residence time was carried out. Figure 6.20 shows that the most prominent primary reaction in both models is the conversion of lignin into LD1 (r7) since its flux is higher than that of the parallel reactions, that is the conversion of lignin into phenolics. The flux of r7 is approximately 0.11 g/mL for both models. The formal kinetic rate coefficient is 4.7·10−2 and 3.6·10−2 min−1 respectively for the ethanol and the water model. Similar observations can be made for the conversion of lignin into methoxyphenols (r1). Hence, the primary lignin degradation seems to be slightly faster in ethanol. However, it was demonstrated in Chapter4 that lignin depolymerisation is faster in water if exactly the same conditions and the same type of lignin were chosen. The type of lignin has thus a significant influence on the kinetics of its primary degradation [JNB10]. Major differences between lignin depolymerisation in water on the one hand and ethanol on the other are the increased yield of catechols and solid residue (see Chapter4). Modelling suggests different reaction pathways for the formation of both catechols and char. While in ethanol catechols are exclusively formed via methoxyphenols, in water the more important pathway is via the lump component LD1. The flux of the conversion of LD1 into catechols at 633 K and 131 min is approximately 2.0·10−3 g/mL and thus significantly higher than the flux of the conversion of methoxyphenols into catechols which results to be approximately 5.8·10−3 g/mL. Comparing the formal kinetic rate coefficients of the reaction pathways originating from catechols (k4, k5, k12) the fastest reaction is the conversion of catechols into phenols. Its formal kinetic rate coefficient is 2.7·10−4 min−1, whereas the formal kinetic rate coefficients of the reactions describing the repolymerisa- −6 −1 tion of catechols (k5, k12) are negligible (<10 min ). The fluxes of all three reactions are very small (<10−4 g/mL). The model suggests thus the repolymerisation of catechols in water to be of almost no importance. This is a contradiction to many previous studies which assume the repolymerisation of monoaromatic compounds to be very important for high yields of char [Sai+03; Tak+12]. Modelling of lignin solvolysis in ethanol also suggests the repolymerisation of catechol to be the more significant reaction causing high yields of char. The flux analysis shows much higher fluxes for this reaction in ethanol than in water (see Figure 6.20). It is indispensable to put more

83 6.3. Lignin depolymerisation in water in a batch-reactor

Lignin  MethoxyPhOH Water MethoxyPhOH  Catechols EtOH MethoxyPhOH  PhOH Catechols  PhOH Catechols  Char

PhOH  CO2, CH4

Lignin  LD1

LD1  LD2

LD1  Char

LD1  Catechols

LD1  CO2

Catechols  LD2

H2O,CO  CO2

CO  CO2

LD1 CO 10-4 10-3 10-2 10-1 Flux (g/L)

Figure 6.20: Fluxes of main reactions comprised by the developed model of hydrothermal depolymerisation of spruce-lignin (see Figure 6.17), reactions r1, r2, r4, r5, r6, r7, r9, r11 are compared with analogue reactions of lignin solvolysis in ethanol (see Section 6.1); τ = 131 min; T = 633 K; lignin/water=132 g/L. eﬀort into the investigation on the key intermediate component(s) which are crucial for the formation of catechols and char since it assists in the com- prehension of the chemical mechanisms. The critical question is whether the inhibition of the repolymerisation of phenolics or the provocation of the cleavage of oligomeric aromates is the main instrument to raise the overall yield of phenolic products.

The formation of gas via LD1 is very similar in both solvents since the fluxes are in the same order of magnitude (see Figure 6.20). The water model considers two reactions that contribute to the gasification of LD1, that are the conversion of LD1 into stable gases (r11), that are CO2 and CH4, and into intermediates (r15), that are CO and H2. The sum of the fluxes of both reactions can be compared with the flux of the gasification reaction of LD calculated by the ethanol model which are 0.02 g/L and 0.01 g/L respectively. The conversion of LD into char and also into gaseous products is suggested to be faster in water than in ethanol. However, it has to be taken into account that the ethanol model deals with a very dominant gas formation reaction

84 Chapter 6. Kinetic studies of lignin degradation which is the gasification of formic acid. The flux of this reaction is almost 10 times higher than the flux of the gasification of LD (see Figure 6.13) which makes it more difficult for the model to describe the latter reaction. The modelling of gaseous and solid products is partially successful. Tendencies can be described, like the rapid decrease of solid residue at short residence times due to the depolymerisation of lignin. However, the underestimation of the gas yield at short residence times and the overestimation at long residence times lead to an increased standard deviation of calculated and measured quantities (19 %). Hence, the modelling of gaseous components has to be reviewed and the model has to be refined. A suggestion for future models is to consider the gasification of lignin as a primary reaction. Obviously the formation of CO2 and CH4 is more immediate. This is also suggested by Yong and Matsumura [YM12].

Figure 6.21: Sensitivity matrices considering comparable reactions and intermediates or products of a) the model of hydrothermal degradation of spruce- lignin and b) the solvolysis in ethanol and formic acid of wheat-straw-lignin; τ = 131 min; T = 633 K; lignin/solvent=132 g/L.

The sensitivity analysis shows very similar results in respect to the reactions discussed above (see Figure 6.21). The high sensitivity of lignin and the moderate sensitivity of all phenolic products to the primary conversion of lignin into LD1 are similar in both solvents. As already shown for the

85 6.3. Lignin depolymerisation in water in a batch-reactor lignin solvolysis in ethanol, also in water the key of increasing the phenolic yield is to suppress the formation of LD and increase the reaction rate of the formation of monomeric methoxyphenols. However, hydrothermal lignin degradations shows that promoting the formation of catechols from intermediates (LD1) must also aid in increasing the yield of phenolics. Char shows a higher sensitivity to the formation of LD1 in water. On the contrary, in ethanol, it shows a higher sensitivity to the formation of catechols. Gaseous products show a moderate sensitivity to the formation and gasification of LD1, whereas in ethanol the sensitivity to those reactions is negligible. This is in good agreement with the conclusions that are drawn from the flux analysis. The low sensitivity of the gaseous products within the ethanol model to the conversion of LD1 into char and gas is certainly due to the dominant gasification of formic acid.

Degradation of lignin The presented model is validated by comparison with literature values. Ta- ble 6.8 shows the apparent kinetic parameters resulting from the kinetic studies of different approaches of thermal lignin degradation, all based on lump-models. Summarised are the most recent works based on pyrolysis, solvolysis and hydrothermal degradation. The kinetic studies of pyrolytic lignin degradation use thermogravimetric analysis (TGA). Some other studies of the last three decades will be additionally discussed in the text. Várh- egyi et al. [Var+97] reported lower apparent activation energies for pyrolytic lignin degradation (34-65 kJ/mol) than Cho et al. [Cho+12] (74 kJ/mol) or Jiang, Nowakowski, and Bridgwater [JNB10] (134-172 kJ/mol) since they used wood without separation of the lignin fraction. Várhegyi et al. conclude that the determination of the kinetics of lignin degradation is difficult since the flat lignin derivative thermogravimetry (DTG)-peak is superposed by the more prominent cellulose peak. A study from 1985 by Nunn et al. [Nun+85] reports activation energies very similar to those found by Cho et al. Jiang, Nowakowski, and Bridgwater [JNB10] analysed different types of lignin and found that the activation energy depends on both, the plant species from which the lignin is separated and the separation method. The apparent Ar- rhenius factor, on the contrary, is less dependent on these two parameters. Jiang, Nowakowski, and Bridgwater additionally considered the reaction order. All other authors considered pseudo-first-order reactions. Studies of the hydrothermal degradation of lignin are mostly based on experiments conducted in batch reactors. Takami et al. [Tak+12] applied a Monte Carlo simulation on the results from gel permeation chromatography (GPC) for the kinetic studies. They found formal kinetic coefficients similar to those

86 Chapter 6. Kinetic studies of lignin degradation obtained in the water model and the ethanol model [Gas+2]. However, all three studies used different types of lignin. The formal kinetic rate coefficients k(633 K) obtained from the works of Zhang, Huang, and Ramaswamy [ZHR11] and Yong and Matsumura [YM12] are significantly higher than those obtained from other studies of lignin depolymerisation. It is worth to mention that the latter study was carried out in a continuous reactor which might have an influence on the heating rate. However, the apparent activation energy documented by Yong and Matsumura is very similar to the apparent activation energy obtained in this work [YM12]. In general the apparent activation energy obtained for pyrolytic lignin degradation tends to be higher than for depolymerisation in water. The resulting formal kinetic rate coefficients obtained for hydrothermal and solvolytic study of lignin degradation spread over a wide range of values. Hence, the reaction rate of primary lignin degradation potentially depends, apart from the thermal degradation method, on the type of lignin and the separation method, on the type of the solvent and the heating rate. Comparing different kinetic studies of lignin degradation based on lump models, it was found that all discussed parameters showed a relevant impact and cannot be discarded. The formal kinetic rate coefficients calculated in this work are in good agreement with the results from other kinetic studies. The thermal degradation of lignin comprises many reactions and complex mechanisms. Efforts concerning the study of detailed kinetics and mechanisms focus on the characterisation of the most abundant chemical structures in the heterogeneous polymer and the linkages between them [Far+10; WR12]. If differences between pyrolysis, solvolysis and hydrothermal processes need to be considered, phase exchange mechanisms, solubility and evaporation conditions of lignin and its degradation products might also have an influence. Reaction mechanisms which are important for the cleavage of C-O bonds or repolymerisation of lignin fragments, especially the differences of pyrolysis and hydrolysis in respect to these mechanisms, are discussed in the following subsections by means of phenolic model substances.

87 6.3. Lignin depolymerisation in water in a batch-reactor a log(A/A’) A 74 5.9 79 2.9 34 3.4 32 1.2 n.a. n.a. n.a. n.a. kJ/mol 134-172 10-13 - b 1 2 1 2 2 2 b − − − − − 2 0 − 1 − 10 10 10 10 10 10 10 10 · · · · · · · · 2.3 in the water model (Figure 6.17 ) 7 k -range k(633 K) E + 1 523-673 6.1 773-1073 6.8 633 4.8 663-733 2.0 573-683 2.0 T 593-653 3.6 K min Organosolv- andhydrolysis-lignin from maple enzymatic by different separationods meth- Lignin fromweak wheat hydrolysis, 9 straw wt% by Organosolv-lignin, 2.3 wt% 623-693 3.0 Alkali-lignin0.1 wt% from spruce, Alkali-lignin10 wt% from Pine, Lignin frommatic spruce hydrolysis, 11.5 by wt% enzy- -Cresol, batch p pyroprobe (batch) Solvolysis in , / FA EtOH batch ter and SCW , continuous NCW , batch DepolymerisationNCW , batch in in the ethanol model (Figure 6.7 ). 10 k + 1 6 − ) k 10 + k 1 min + ) 6 7 k k = 1 0 + + 1 1 extrapolated rate coefficient A b a [ Cho+12 ] Pyrolysis, TGA - MS and [ JNB10 ] Pyrolysis, TGA Several types of lignin obtained Ethanol-model (k ][ Tak+12 Depolymerisation in wa- [ YM12 ] Depolymerisation in [ ZHR11 ] Depolymerisation in SourceWater-model (k Thermal treatment Lignin, Concentration and by k Table 6.8: Comparison of differentobtained kinetic from studies of this thermal work lignin refer degradation, to all based a on lump lump-models; of values the reactions represented by k

88 Chapter 6. Kinetic studies of lignin degradation

Decomposition of methoxyphenols

The validation of the present model was thus additionally supported by the comparison of formal kinetic parameters of the decomposition of homogeneous reactions of less complex components occurring within the thermal degradation of lignin. Table 6.9 shows the results of different kinetic studies of the thermal decomposition of methoxyphenols and catechols. Methoxyphe- nols tend to decompose by the cleavage of the methyl C-O bond. The aromatic ring and the hydroxyl group are more stable [Vuo86]. The formal kinetic rate coefficients at 633 K k(633 K) obtained in the water model is 1.1·10−2 min−1 which is in good agreement to other kinetic studies of hydrothermal and solvothermal decomposition of methoxyphenols [For+12; Gas+2; WSG11]. Values of k(633 K) obtained by studies of pyrolytic decomposition of methoxyphenols are lower and range from 2.8·10−5 min−1 to 2.1·10−3 min−1 [DM99; KV08; Vuo86]. However, the activation energy obtained from pyrolysis is significantly higher than that obtained from hydrothermal lignin degradation in NCW and SCW. Dorrestijn and Mulder [DM99] claimed that guaiacol decomposes via a radical-induced mechanism. They studied the kinetics of guaiacol decomposition at elevated temperatures. The obtained kinetic parameters are in good agreement of those obtained by Vuori [Vuo86] who also studied pyrolytic decomposition of guaiacols at lower temperatures. The latter author reported, that in addition to free radical chain reactions hydrolytic decomposition of guaiacols occurs. He claims that this is due to the presence of water which is formed during the pyrolytic decomposition of guaiacol. Furthermore, Lawson and Klein found that the formal kinetic rate coefficient at 656 K slightly increased if water was present [LK85] which indicates a shift of the dominant reaction mechanism from radical-induced to hydrolytic cleavage. Since water is abundantly present, hydrolysis is assumed to be the dominant mechanism for the hydrothermal decomposition of methoxyphenols [TRR11]. In a previous work was discovered that around the critical point of water a shift of the apparent activation energy of guaiacol decomposition occurs due to a change of the dominant reaction mechanism from hydrolytic to radical-induced decomposition [For+12]. This explains that the apparent activation energy obtained from different studies conducted in NCW or SCW vary in a relatively wide range from 39 to 85 kJ/mol [For+12; WSG11]. The apparent activation energy obtained from studies of thermal decomposition of methoxyphenols in organic solvents or without solvent are significantly higher (> 170 kJ/mol). Due to the lower activation energy hydrolytic cleavage of the C-O bond of methoxyphenols is faster at moderate temperatures (< 673 K).

89 6.3. Lignin depolymerisation in water in a batch-reactor

Decomposition of catechols Furthermore, kinetic studies carried out in order to determine the formal kinetic rate coefficients of the thermal decomposition of catechol shall be discussed. Catechol decomposes into other phenolic substances, e.g. phenol [Nim+11a]. In addition it is potentially responsible for the formation of char by repolymerisation [McM+04]. Wahyudiono, Sasaki, and Goto studied the hydrothermal decomposition of catechol and found higher molecular structures. Their amount increases with residence time and temperature. However, the formation of phenol and other monoaromatic components was dominant [WSG09]. Analogously modelling of the hydrothermal lignin degradation shows, that the polymerisation of catechol is of minor importance since the competing reaction, which is the conversion of catechols into phenols (r4), is faster (see Table 6.7). The study of lignin degradation kinetics by Yong and Matsumura [YM12] reports a higher formal kinetic rate coefficient for polymerisation than for the conversion into phenol. However, the lumped formal kinetic coefficient comprising conversion into phenols and polymerisation obtained by Yong and Matsumura (3.9 min−1) differs significantly from the values obtained in the water model (2.9·10−4 min−1) and by Wahyudiono, Sasaki, and Goto 2.4·10−4 min−1.

90 Chapter 6. Kinetic studies of lignin degradation a log(A/A’) A 39 1.5 51 0.6 85 5.0 55 1 170239 12 237190 15 15 151 13 12 n.a. n.a. kJ/mol b b b 2 2 2 3 5 5 3 4 4 b − − − − − − − − 0 − 1 − 10 10 10 10 10 10 10 10 10 10 · · · · · · · · · · in the ethanol-model (Figure 6.7 ), 2 -range k(633 K) E T 593-653 1.1 603-633 1.5 633 8.7 593-653 2.9 663-733 3.9 K min in the water model. 12 k + 5 k + 4 matic hydrolysis, 11.5 wt% Lignin frommatic spruce hydrolysis, 9 by wt%, Lignin enzy- fromweak wheat hydrolysis, 9 straw wt% by matic hydrolysis, 11.5 wt% Guaiacol, 1.24 wt%Guaiacol, 653-673 6 mol% 1.8 Alkali-lignin 688-7840.1 wt% from 3.0 spruce, in the water-model (Figure 6.17 ) and by k and cumene, 3 2 k + 2 Solvolysis in NCW , batch Lignin from spruce byNCW , enzy- batch Solvolysis in ethanol/formic acid, batch Solvolysis in ,NCW batch Lignin from spruce by enzy- continuous SCW , batch ous 1 − ) 12 k min ) + 3 5 k k = 1 0 + ) + 2 2 4 extrapolated rate coeﬃcient A b a SourceWater-model (k Thermal treatment][ For+12 Ethanol model (k Reactant, Concentration Catalysed[ Dor+99 ] solvolysis][ Vuo86 in [ KV08 ] Pyrolysis in N Water-model (k Pyrolysis in tetralin, batch Pyrolysis, batch[ WSG09 ] Guaiacol, 50 mol% Solvolysis in NCW , batch Guaiacol Catechol, 598-648 1.1 wt% 2.8 643-663 2.4 523-800 2.1 [ WSG11 ] Solvolysis in NCW and [ YM12 ] Solvolysis in SCW , continu- and catechols, lump of the reactions represented by k Table 6.9: Comparisonreactions of represented diﬀerent by kinetic k studies of thermal decomposition of methoxyphenols, lump of the

91 6.3. Lignin depolymerisation in water in a batch-reactor

92 Chapter 7

Feasibility study of lignin liquefaction

A preliminary feasibility study is essential for the evaluation of a technical process. Further research should follow a pathway which is economically promising. The intention in this chapter is to compare two approaches of lignin liquefaction, solvolysis in ethanol and hydrothermal degradation, in order to discover which one provides a higher economic potential. The assumption of the parameters essential for a technical realisation of the production of phenolics from lignin leads to a rough estimation of the total manufacturing costs (COM) of phenol. A preliminary design of the technical process including the main apparatus and equipment was carried out in order to estimate the module costs and the total capital investment (TCI) of the plant. For this utility and labor costs had to be estimated. Exper- imental results from Exp. 66 and 98 were employed to estimate the input amount of lignin and solvent as well as the yields of phenolic, gaseous and solid products.

7.1 Background Scenario

A technical process consists of the following units: pre-treatment, reactor, separation and puriﬁcation. For this study was assumed that the lignin conversion unit is a part of a larger facility, treating the lignocellulosic biomass as a whole. Hence, the pre-treatment is part of the existing global infras- tructure. In the case of solvolysis in ethanol it can be assumed that the plant is integrated in an ethanol-production plant comprising the fermentation of

93 7.2. Process design cellulose for ethanol production and the liquefaction of lignin in ethanol. In order to calculate the streams and ﬂuxes within the process the capacity of the plant has to be deﬁned considering a global biomass treatment plant including both the fermentation of the lignocellulosic raw material and the liquefaction of lignin in ethanol. Humbird et al. [Hum+1] carried out a process design of a fermentation plant with a capacity of 730 kt of corn stover per year. The design of a plant converting softwood into ethanol by Wingren et al. [WGZ03] deals with a capacity of 200 kt of dry input material per year and about 60000 m3 ethanol per year. Haase [Haa2] carried out a cost estimation for a plant with a capacity of 450 kt of dry wood per year. The hydrothermal lignin degradation might be integrated into a pulp mill. The dry matter of the residue from the pulp production, so called black liquor, contains up to 50 % lignin. Black liquor is the main waste stream emerging from the pulp process. It is dried and burned in order to recover the energy from the organic material and to recycle the chemicals for the pulp process. The integration of the lignin liquefaction into a pulp process would provide the most convenient feedstock situation since phenolics would be produced from grave to cradle by using an internal waste stream as feedstock [Tow+07]. The energy and cost intensive drying of the black liquor would be eliminated. Furthermore, the char which is produced within the hydrothermal treatment of the lignin could still be used for the generation of heat or electric power. However, the behaviour of the chemicals in a hydrothermal process and their recovery for further usage in the pulp mill has not been studied so far. Thus, and for a better comparison of the hydrothermal process with the ethanol approach, the conditions of the hydrothermal process were chosen to be equal to those of the ethanol process. For this study, the size of the plant was chosen to be rather small compared to other scenarios mentioned above due to the great amount of solvent in the feed and the relatively long residence time (up to 90 min) in the reactor. For the following calculations the capacity of the plant was assumed to be 8000 t/a of dried biomass. The plant was assumed to run 8000 h/a.

7.2 Process design

For the solvolytic lignin degradation a CSTR was assumed to be the most eﬀective reactor based on the results documented in Chapter 6.2. The input stream consisting of lignin and fresh solvent is ﬁrst mixed with the recycled solvent and thereafter pressurised by a pump (A). Before entering the reactor (C) it is preheated by the output stream from the reactor in a heat exchanger

94 Chapter 7. Feasibility study of lignin liquefaction

(B). The reactor is heated by a thermal oil heater (D). The thermal oil is heated up in a gas-combustion chamber. The advantage of this indirect heating system is that all tubes in the heating-circle are low-pressure tubes. In addition, it ensures the prevention of explosion or ignition in case of a leak in the feeding tube. A direct heating of the feed in a combustion chamber would be too dangerous, especially if it contains ethanol and formic acid. If water is used as solvent, a direct heating of the input stream might be an economically more interesting solution. However, in the case of direct heating the tubes which conduct the input stream through the combustion chamber have to be designed to withstand the high pressures of the fluid. The output stream is cooled in the first heat exchanger (B) by the input stream. The solid particles, that are char or not reacted lignin, can be separated by e.g. a cyclone (E) before the output stream is further cooled in a second heat exchanger (F) by fresh water. After separating the gases in e.g. a flash separator (G), the solvent is separated from the product and recycled. The product is further separated into phenol, other phenolics and a fraction consisting of all other organic substances which usually have a tarry or oily texture (H). A detailed design of the separation process was not carried out.

Lignin 1 P-78 Solvent

2 10 11 Co-Solvent Purge Product Gas 3 A Air Flue Gas 6

8 G Phenol 4

Water 9 Phenolics F H D Steam 7 Oil/Tar C B E Natural Gas 5 Char Product Gas

Figure 7.1: Process scheme with main components (labelled with letters) an streams (labelled with numbers).

95 7.3. Main equipment

Table 7.1: Streams deﬁned for the lignin liquefaction process (see Figure 7.1) for both, ethanol and water approach. Stream 1 2 3 4 5 6 Unit t/h t/h t/h t/h t/h t/h EtOH 1.00 1.34 1.22 11.70 0.05 1.11

H2O 1.00 0.50 0.00 8.58 0.34 0.18 Stream 7 8 9 10 11 Unit t/h kg/h kg/h t/h t/h EtOH 2.35 3.00 63.00 8.14 0.04

H2O 0.44 1.40 31.20 7.08 0.50

7.3 Main equipment

For a preliminary cost estimation the necessary equipment items have to be identified and their costs have to be estimated. Based on the process displayed in Fig. 7.1 the main equipment items are the pump, the heat exchangers, the jacketed stirred reactor, the heater including combustion chamber, chimney, thermal oil and circulation pump. The costs of the equipment necessary for the separation of the products is estimated by a factor. The dimensions of the different apparatus are estimated using the results from the experiments Exp. 66 and 98 for the ethanol and water approach respectively. The price increase for offers collected earlier than 2012 was estimated by the price-index by Kölbel/Schulze [VCI12]. The conversion of the currency was estimated by the Euro-course from September 2012 (1.29 $/ e).

Pump The selection of the pump for the transfer of the feed suspension is essential for the estimation of its costs. The pumped fluid consists of the liquid solvent and solid lignin particles. The target pressure was 25 MPa and the temperature of the fluid was assumed to be 343 K. For the selection of the pump the volumetric flow rate of the fluid, that is the capacity V˙ , had to be calculated (see Equation 7.1). Hence, the mean density of the pumped fluid ρ was estimated by considering the density of all constituents i of the fluid at 343 K. The density of ethanol, formic acid and water at 343 K are 0.74, 1.16 and 0.98 t/m3 respectively [VDI06]. The density of lignin was estimated to be 1.4 t/m3 [Haa2]. The most convenient pump for the given transfer challenge is a plunger pump (displacement pump). The total dynamic head H was calculated employing Equation 7.2 considering the gravitational acceleration

96 Chapter 7. Feasibility study of lignin liquefaction

g. The useful work Wo was calculated employing Equation 7.3 considering the pressure difference between the initial pressure pin and the setup pressure in the reactor pout. The influence of valves and accessories or tubes on the pressure causing a loss of head was neglected. The efficiency of the pump was assumed to be 85 %.

!−1 m˙ X m˙ i · ρi V˙ = =m ˙ (7.1) ρ m˙ i p − p H = in out , g = 9.81 m/s2 (7.2) ρ¯ · g

Wo = V˙ · (pin − pout) (7.3) All parameters necessary for the estimation of the purchased costs of the plunger pump are listed in Table 7.2. The purchased costs for the pump can be estimated using the data from Peters [PT04, p.518]. According to Peters the purchased costs for the pump range from 6660 e for the water approach to 9030 e for the ethanol approach. However, according to the oﬀer from Schäfer & Urbach for a plunger pump with a capacity of 16.8 m3/h and 4100 m total dynamic head in 2010 the purchased costs are 108 k e (see AppendixF), 120 k e considering the price-index for 2012. For this study the costs were estimated based on the oﬀer from Schäfer & Urbach for both the ethanol and the water approach, which is the more conservative estimation.

Table 7.2: Parameters used for the estimation of the purchased costs of the pump. Pump (A) ρ m˙ V˙HW o W kg/m3 t/h m3/h m kW kW EtOH 844 11.7 13.9 3020 113 133 Water 1027 8.6 8.4 2482 58 68

Reactor In respect to the investment costs the reactor is commonly the less expensive part of a chemical plant. Separation and puriﬁcation units form the major part of these costs. The costs of the reactor were estimated by its size. Assuming a capacity 1 t/h dry lignin, the total throughput passing the reactor is 11.7 t/h for solvolysis in ethanol and formic acid and 9.58 t/h for the hydrothermal approach. The density in the reactor was estimated

97 7.3. Main equipment by the density of the solvent at the determined conditions (T and p) in the reactor. In Table 7.3 the reaction conditions are listed for both ethanol and water approach. The mean residence time τ 0 was estimated according to the experimental results. The resulting reactor volume VR was calculated by the mass ﬂow m˙ (see Section 7.1), the density according to Bazaev et al. [Baz+07] and the mean residence time. The purchased costs of the jacketed stirred reactor in dependence of the reactor volume were estimated according to Peters [PT04, p.628]. The increased costs for a high-pressure reactor are considered by multiplication of the purchased costs of a none pressurised reactor with the factor 10.6 [PT04, p.556]. The data collected from Peters are dated for the year 2002. The resulting factor for the price increase is 1.28 [VCI12]. The purchased costs of the reactor for the ethanol and the water approach were 1.482 M e and 1.270 M e respectively.

Table 7.3: Reaction conditions and resulting reactor volume VR for both ethanol and water approach. 0 Approach Tpτ ρ m˙ V R K MPa min kg/m3 t/h m3 EtOH 673 25 44 280 11.7 30.5 Water 653 25 90 446 8.6 28.8

Heat exchanger In order to calculate the purchased costs of the two heat exchangers (B and F), the heat transfer area AEx had to be determined (see Equation 7.4). Both heat exchangers were assumed to be counter current ﬂow heat exchanger with 2 a transfer coeﬃcient ksum of 600 W/(m ·K) [Wag09]. Heat exchanger (B) was employed to pre-heat the input stream from a temperature below the evaporation temperature of the solvent (Tin = 343 K) to a temperature of Tout = 453 K and to cool the product stream entering the heat exchanger with a temperature of 673 and 653 K in the ethanol and the water approach, respectively. In the heat exchanger (F) the product stream requires further cooling to a temperature of Tout = 343 K by an independent water stream (see Fig. 7.1) which was suggested to be evaporated at atmospheric pressure (0.1 MPa). The steam leaving the heat exchanger (F) should have a temperature of around Tout = 453 K in order to be used elsewhere in the plant as a source of thermal energy. q˙ AEx = (7.4) ksum · ∆T

98 Chapter 7. Feasibility study of lignin liquefaction

Table 7.4: Data concerning heat exchanger (B and F) for both ethanol and water approach. Heat- Cold Stream Hot Stream

Exchanger Tin Tout cp Tin Tout cp ∆T q˙A Ex kJ kJ 2 KK kg·K KK kg·K K MW m EtOH B 343 453 3.40 673 575 3.80 226 1.22 9.0 EtOH F 298 453 4.18 583 343 3.65 80 2.84 68.1 Water B 343 453 4.19 653 584 6.71 220 1.50 11.4 Water F 298 453 4.18 584 343 4.47 81 3.51 72.5

For the determination of AEx the rate of heat transfer q˙ was calculated by Equation 7.5.

q˙ = cp · ∆T · m˙ (7.5)

The specific heat capacity was estimated by the average heat capacity cp of the solvent for the corresponding temperature difference ∆T . In order to determine ∆T the entering and emerging temperature of one stream, that is either hot or cold stream, had to be defined. These temperatures were partially determined through the reaction parameters of the model experiments and partially through reasonable assumptions (see explanation above and Table 7.4). Furthermore, for the calculation of AEx the mean overall temperature difference between the two fluids ∆T has to be determined according to Equation 7.6

∆Tin − ∆Tout ∆T = , (7.6) ln ∆Tin ∆Tout where the difference between the temperature of the hot fluid emerging the heat exchanger and the cold fluid entering the heat exchanger is ∆Tin and the difference between the temperature of the hot fluid entering the heat exchanger and the cold fluid emerging the heat exchanger is ∆Tout. The rate of heat transfer in both heat exchangers (B and F), the input temperatures of cold and hot stream, the mean overall temperature difference as well as the heat transfer area are listed in Table 7.4. After calculation of the heat transfer area AEx the purchased costs for the heat exchangers could be estimated according to Peters [PT04, p.681]. Considering also the price-index and the dollar-course the purchased costs of the heat exchangers for the ethanol and the water approach are 25 k e and 37 k e respectively.

99 7.4. Cost estimation

Thermal oil heater For the estimation of the purchased costs of the thermal oil heater, the rate of heat transferred to the feed was calculated with Equation 7.5 analogously to the transferred heat in a heat exchanger. The initial lower temperature of 453 K is equal to the output temperature of the cold stream of heat exchanger (B). The target temperature is equal to the reaction temperature, which is 673 K and 653 K for the ethanol and water approach respectively. The mean specific heat capacity (cp) of the fluid of the feed was determined as shown in Equation 7.1 and resulted to be 3.8 and 5.4 kJ/(kg·K) for the ethanol and the water approach respectively. The resulting rate of heat transfer q˙ was 2.72 and 2.57 MW for the ethanol and water approach, respectively. Assuming an efficiency factor of 0.85, the power capacity of the heater was calculated to be at least 3.2 MW and 3.1 MW for the ethanol and water approach, respectively. The basis for the cost estimation of the heater was an offer from HTT for a thermal oil heater from 2010 with a power capacity of 3.4 MW. The offer is supposed to be suitable for a conservative cost estimation. The price-index by Kölbel/Schulze indicated a price increase factor from 2010 to 2012 of 1.06. The resulting purchased costs for both ethanol and water approach is 617 k e.

7.4 Cost estimation 7.4.1 Market prices For the following calculations data about material and utility prices, prices for labor, etc. have to be provided. The required data is collected from diﬀerent sources in order to estimate the costs for both the ethanol and the water approach, and is presented in Table 7.5. The price of lignin was estimated considering the prices of alkali-lignin which is a waste material and thus ﬁts best into the background scenario. The prices for alkali-lignin vary in dependence of the degree of drying [Pul09]. For the ethanol approach dried lignin should be used, for the water approach a lignin containing 50 % moisture or more is also a possible feedstock. Market prices for technical formic acid with 85 % purity were considered for this study although for the experiments formic acid with a higher purity was used. However, prices for bulk amounts of formic acid with a higher purity were not available. The price of 390 e/t for ethanol can be achieved in Brazil, in Europe the price is up to 500 e/t [Haa2]. The price of waste water (purge stream 11) which is produced in large quantities in the water approach was estimated according to Peters who reports prices of process

100 Chapter 7. Feasibility study of lignin liquefaction

Table 7.5: Estimated prices for reactants, products and utilities. Input/Output material Quality Bulk Prices Unit Source Alkali-lignin dried 480 e/t [EIA11] Alkali-lignina 50 % 160 e/t [Pul09] Formic acid 85 % 800 e/t [Imh12] Ethanol 96 % 390 e/t [Haa2] Water 1 e/t [PT04] Phenol 1600 e/t AppendixF Other phenolics 953 e/t AppendixF Natural Gas 0.048 e/kWh [VCI11] Product-gas EtOH 136 e/t [VCI11] Product-gas Water 21 e/t [VCI11] Char 107 e/t [VCI11] Tar/Oil 12.5 % 49 e/t [VCI11] Electricity 0.13 e/kWh [VCI11] Steam 0.03 e/kWh [VCI11] a Lignin contains 50 % water, price refers to dry lignin. water of maximum 0.46 $/t [PT04, p.898]. Considering a price increase from 2002 to 2012, the price was assumed to be signiﬁcantly higher (see Table 7.5). The price of phenol was estimated using the price report by ICIS in July 2012. The prices of other phenolics, that are catechols and methoxyphenols, were estimated with the benzene-price which is usually the raw material for the production of these substances. Both the ethanol and the water approach yield considerable amounts of gas, char and oil. These products can be used for the generation of heat which causes a reduction of the material or the utility costs. The bene- ﬁt or credit however depends on the heating value of the products. The lower heating value (LHV) of the gaseous products can be determined by calculating the LHV of its constituents. The gaseous products from the ethanol experiment have a LHV of 4.2 kWh/kg, those from the water experiment 0.65 kWh/kg (see Appendix E.2). Assuming a price for natural gas of 0.327 e/kWh [VCI11] the credit which can be achieved for the gaseous products of the ethanol and the water approach were estimated to 136 and 21 e/t respectively. The higher heating value (HHV) of the char was estimated by the results from the elemental analysis and employing the calculation method developed by IGT [Tal81]. The HHV of the solid residue from both the ethanol experiment Exp. 66 and water experiment Exp. 98 are 31 MJ/kg (see Appendix E.2). This value is in the range of the HHV of high bituminous coal which has a HHV above 30 MJ/kg [DIN84]. Thus the price

101 7.4. Cost estimation for char was estimated by the price for coal which is 107 e/t in 2011 [VCI11]. The oil-fraction is comprised of all liquid products excluding the solvent and the phenolic products. The LHV of the oil-fraction can be estimated only for the ethanol experiment since the obtained results for this case allowed to carry out an energy balance (see Appendix E.2). The resulting LHV was 5.3 MJ/kg which is about 12.5 % of the minimum LHV of fuel oil (42 MJ/kg [DIN84]). Thus the credit for the oil-fraction was estimated to be 12.5 % of the fuel oil price in 2011 (see Table 7.5). Although the LHV of the oil fraction obtained from the water experiment could not be determined, the estimation of the credit was assumed to be valid for both the ethanol and the water approach.

7.4.2 Manufacturing costs For the preliminary cost estimation diﬀerent cost items were considered, that are the TCI, the total material costs (TMC), total utility costs (TUC) and the operating labor costs (OLC). Based on these cost items the COM of the product (phenol) were estimated. This value serves best to carry out an economic evaluation of the process.

Investment costs (TCI) The TCI are based on the costs of the main equipment, which were determined in the preceding section. In addition, costs for the separation equipment were estimated in order to determine the total bare module costs (TBM). Costs for additional equipment like tubes, valves, control units, buildings, engineering and construction were estimated by applying empiric factors (see Table 7.6). These factors were taken from Baerns [Bae+06, pp 463-477].

Utility costs (TUC) Utility costs were calculated on the basis of experimental results. Among all input and output streams (see Table 7.1) the following streams were suggested to be essential for the calculation of the TUC since their utilisation was foreseen for the purpose of energy generation: The output of gaseous (6) and solid (5) products and the output of liquid (7) products (oil/tar) excluding phenol (8) and phenolics (9). The necessary electric and thermal energy is estimated from the energy balance for experiment Exp. 66 (see Appendix E.2) and the required energies for the heater and the pump calculated in Section 7.3. The steam recovered from heat exchanger (F) was

102 Chapter 7. Feasibility study of lignin liquefaction

Table 7.6: Total Capital Investment TCI calculated by various factors based on TBM.

Item Factor Costs

EtOH-approach H2O-approach M e M e Heater 0.617 0.617 Reactor 1.482 1.270 Pump 0.120 0.120 Heat exchanger 0.027 0.040 Separation equipment 0.70 5.241 4.776 TBM 1.00 7.487 6.823 On-Site Costs Equipment installation 0.15 1.123 1.024 Tubes & Valves 0.80 5.989 5.459 Measuring and control engi- 0.35 2.620 2.388 neering Electrical equipment 0.20 1.497 1.365 Buildings & site preparation 0.65 4.866 4.435 Insulations & ﬁre protection 0.15 1.123 1.024 Oﬀ-Site Costs Engineering 0.40 2.995 2.729 Unexpected costs 0.20 1.497 1.365 Capital Costs (TCI) 3.90 29.198 26.611

assumed to be used as energy source elsewhere in the process. A surplus of 50 % of the required energy was considered for the separation equipment. The total utility costs are displayed in Table 7.7.

Operating labor (OLC) The operating labor costs for a common bio reﬁnery which runs 8000 h/a were analysed by Baerns [Bae+06, pp 463-477]. The operation of the plant requires 2 skilled workers and one supervisor for each of the ﬁve shifts. In addition, one technician and one head of unit are required. The total operating labor costs for the entire plant including ethanol production and lignin liquefaction are thus 1.172 M e/a. The lignin production is estimated to require 1/3 of the operating labor of the entire plant which is calculated to 0.391 M e/a (see Tab. 7.8) for both ethanol and water approach.

103 7.4. Cost estimation

Table 7.7: Energy demands and utility costs (TUC).

Utilities EtOH-approach H2O-approach Demand Costs Demand Costs MWh/a M e/a MWh/a M e/a Electricity 912 0.114 912 0.114 Natural gas 27200 1.295 25600 1.219 Separation (overall 50%) 0.726 0.617 Steam -22745 -0.682 -23840 -0.715 TMC 1.453 1.235

Table 7.8: Operating Labor Costs (OLC).

Wages Annual Costs Rel. Costs

EtOH-approach H2O-approach k e/a M e/a e/kgP he) e/kgP he) Skilled workers 64 0.640 26.67 57.14 Supervisor 70 0.350 14.58 31.25 Technician 72 0.072 3.00 6.43 Unit head 110 0.110 4.58 9.82 OLC bio reﬁn- 1.172 48.83 104.64 ery OLC phenol- 0.391 16.28 34.88 production

104 Chapter 7. Feasibility study of lignin liquefaction

Total manufacturing costs COM For the determination of the COM the annual capital related costs which comprise depreciation, maintenance, working capital, contingency and fees were assumed to make up 23 % of the TCI. The maintenance and working capital were estimated similar to a fossil fuel reﬁnery and lower than a chemical plant. The depreciation was assumed to be linear and distributed over ten years. The residual value after ten years of the plant was assumed to be equal to the disassembly costs [Bae+06, pp 463-477]. Furthermore, costs for transport, laboratory charges, packaging and distribution as well as waste treatment were considered. The sum of these costs was estimated to be around 2.9 % of the overall manufacturing costs and is listed in Ta- ble 7.9 under miscellaneous. The relatively low value is a consequence of the assumption that the plant is part of a larger facility. The resulting manufacturing costs are 909 e and 749 e per kilogram of produced phenol for the ethanol and water approach, respectively. This is several hundred times higher than the actual market price of phenol (1.6 e/kg).

7.5 Discussion

The preliminary cost estimation shows that the COM (see Table 7.9) for phenol from lignin are much higher than the market price of phenol (1.6 e/kg). Both the ethanol and the water approach are thus currently economically not feasible, although the water approach shows lower manufacturing costs at still very low yields of phenol (0.14 wt%). An increase of the yield can reduce the manufacturing costs dramatically. In order to find the bottleneck- parameters which mainly influence the COM, different scenarios were observed. The parameters which were contemplated within this work were:

• the phenol yield; depending on the type of the lignin used and the separation method applied on the lignin, the yield of phenolics and especially phenol can be increased to maximum about 30 wt% [APR01]. However, the maximum yield of phenol from waste lignin containing 50 to 70 % of cellulose and impurities will be signiﬁcantly smaller. 10 to 20 % seem realistic. Thus the inﬂuence of the variation of the phenol yield on the COM should be investigated.

• The lignin/solvent ratio in the feed. This parameter might be increased to 30 wt%. Black liquor containing 30 wt% dry material is still pumpable since most of the biomass is diluted. The transfer of an ethanol/lignin suspension is assumed to be more challenging if the

105 7.5. Discussion

Table 7.9: Total Manufacturing Costs COM.

Costs EtOH-approach H2O-approach Annual Costs Rel. Costs Annual Costs Rel. Costs M e/a e/kgP he)M e/a e/kgP he) Raw Material Lignin 3.84 160.00 1.28 114.29 Solvent 4.19 174.59 0.00 0.36 Formic Acid 7.81 325.33 0.00 0.00 Credits Phenolics -0.48 -20.01 -0.24 -21.24 Product Gas -1.78 -74.24 -0.05 -4.04 Char -0.03 -1.39 -0.30 -27.03 Oil -0.92 -38.31 -0.17 -15.52 Sum material 12.62 525.98 0.52 46.82 Utilities 1.45 60.53 1.23 110.24 Operating La- 0.39 16.28 0.39 34.88 bor Capital related 6.72 279.81 6.12 546.48 costs Miscellaneous 0.64 26.48 0.25 22.15 COM 21.82 909.08 8.52 748.76

percentage of solid lignin is increased since the solubility of lignin in ethanol is lower than in basic aqueous media.

• Decreasing the mean residence times will reduce the TCI due to a smaller reactor at a constant capacity of the plant. The values employed to calculate the basic case for the water approach were taken from a batch experiment. However, the results presented in Chap- ter 6.2 showed that in a CSTR the residence time can be reduced signiﬁcantly due to e.g. the elevated heating rate.

• The reduction of the utility costs due to optimisation of the energy ﬂuxes can be achieved by a reduction of the consumption of natural gas. A reduction of 25 % (800 kW) seems reasonable, if e.g. the reaction enthalpy (9600 kW, see Table E.5) calculated for the ethanol approach can be partially recovered.

106 Chapter 7. Feasibility study of lignin liquefaction

• The reduction of the module costs of the heater by 30 % if direct heating of the feed is applied. This option is only possible for the water approach due to safety aspects. • The reduction of the costs of the separation equipment from 70 % to 60% of the TBM. Since no detailed estimation of the module costs for separation equipment was carried out, this factor can vary between 40 and 70 % [Bae+06, pp 463-477]. Obviously these costs depend also on the phenol yield and the selectivity for special phenolic products. It was already shown in Chapter4 that the water approach shows a narrower product spectrum and a high selectivity for catechols. This is supposed to facilitate the separation of the products and thus decrease the module costs. • The lignin price assumed for the basic case is above the costs for the recovery of lignocellulosic biomass [Haa2]. If the lignin is recovered from a waste stream, the prices for lignin can be reduced.

Considering the mentioned parameters three scenarios different to the basic case were suggested (see Table 7.10). The composition of the COM for the basic case of the ethanol approach shows that the material costs are the main contributor to the COM. The reasons thereof are the consumption of formic acid and the gasification of ethanol. On the contrary for the water approach the capital related costs make up the greatest part of the COM (see Figure 7.2). For the design of a plant which deals with high pressures, the TCI are high since reactor, pipes, heater, etc. have to withstand these high pressures. In addition the amount of solvent which circulates in the plant is much higher than the throughput of lignin. This results in big vessels and pipes that are accordingly more expensive. Hence, the COM would shrink significantly if the lignin/solvent ratio was increased to 30 % (Scenario 1). The application of Scenario 1 on the water approach has no influence on the material costs since the yields of phenol, char, gas, etc. were assumed to be constant. However, also the overall capacity of the plant, and the percentage of ethanol which is gasified were assumed to be the same as in the basic case. Thus, in absolute numbers, less formic acid was assumed to be used and less ethanol gasified. This leads to a dramatic decrease of the material costs for the ethanol approach. In Scenario 1 the COM were determined to be lower for the ethanol than for the water approach. If the mean residence time could be reduced, exclusively the capital related costs would shrink (Scenario 2, Figure 7.2). A mean residence time of 10 min reduces the capital related costs to 59 % and the COM to 71 % of the original costs in the basic case for the water approach. For the ethanol

107 7.5. Discussion

a ) b ) M i s c e l l a n e o u s 1 0 0 0 1 0 0 0 O p e r a t i n g L a b o r U t i l i t y 8 0 0 8 0 0 C a p i t a l r e l a t e d )

g M a t e r i a l

k 6 0 0 6 0 0 / € (

s 4 0 0 4 0 0 t s o 2 0 0 2 0 0 C

0 0 B a s i c S c e 1 S c e 2 B a s i c S c e 1 S c e 2

Figure 7.2: COM of phenol from lignin for the basic case and 2 diﬀerent scenarios (see Table 7.10) for a) the water approach and b) the ethanol approach.

1 0 3 C O M E t O H

H 2 O 1 0 2 S c e 3 ) g k /

€ 1

( 1 0

M O 0 C 1 0

1 0 - 1 1 0 - 3 1 0 - 2 1 0 - 1 1 0 0

P h e n o l y i e l d ( g / g L i g n i n )

Figure 7.3: COM of phenol from lignin in dependence of the phenol yield for the basic case of both ethanol and water approach; scenario 3 (see Table 7.10) is displayed for the water approach.

108 Chapter 7. Feasibility study of lignin liquefaction

Table 7.10: Scenarios of the plant design and cost estimation for solvolytic lignin liquefaction and phenol production

Scenario Modiﬁed parameters Approach Aﬀected parameters Sce1 Lignin/solvent 30 wt% EtOH Input lignin 2.7 t/h

H2O Input lignin 2.0 t/h Sce2 τ 10 min EtOH Reactor Vol. 7.0 m3 3 H2O Reactor Vol. 3.2 m

Sce3 Lignin/solvent 30 wt% H2O Input lignin 2.0 t/h τ 10 min Reactor Vol. 3.2 m3 Natural gas 75 % consumption Heating 70 % equipment Separation 60 % equipment Lignin price 107 e/t

approach the COM is reduced to 88 % of the original value in the basic case. The greater reduction for the water approach results from the longer mean residence in the basic case compared to the ethanol approach. Furthermore, since the module costs in the basic case of the water approach are lower than in the basic case of the ethanol approach, the reduction of the COM with decreasing residence times is more dramatic for the water approach. An increased phenol yield reduces the COM most significantly. A yield of 10 % would reduce the COM to 10.70 e and 27.32 e per kilogram phenol for the water and the ethanol approach, respectively (see Figure 7.3). However, this is still too much to make the process economically attractive. And even if the phenol yield could be increased to 30 %, the COM were still above the market price of phenol. Scenario 3 assumes that several parameters can be improved (see Table 7.10). However, even though the residence time could be reduced to 10 min, the module costs for the heater and the separation equipment as well as the consumption of natural gas and the price of lignin would decrease, the phenol yield still has to be increased to 20 % in order to gain a profit from the production of phenol from lignin. The production of phenol from lignin by solvolysis in water or ethanol is economically not feasible if the current yields and process parameters are considered. The most important parameter influencing the feasibility is the phenol yield. However, even if the phenol yield could be increased to 45 wt%

109 7.5. Discussion referring to the input amount of lignin, which seems not very realistic at the current state of research, a process would still not be economically attractive. A reduction of the mean residence time in the reactor, an increase of the lignin/solvent ratio are also key parameters which contribute to the necessary drastic reduction of the capital related costs. In fact this study provides a rough and conservative estimation of the capital related costs. A more detailed knowledge about the equipment costs, especially the separation equipment, might show that these costs can be reduced. The water approach shows better economic opportunities than the ethanol approach since the feedstock situation provides large quantities of cheap feedstock, the solvent is more stable and abundantly available at low prices. Future research should thus focus on black liquor. However, the yield of phenol and other phenolics has to be increased to at least 20 %, the lignin/solvent ratio should be increased signiﬁcantly (20 to 30 %) and the mean residence time has to be reduced to maximum 10 to 30 min in order to make the process economically attractive.

110 Chapter 8

Conclusion

This work deals with the thermal depolymerisation of lignin, especially the solvolysis in ethanol and the hydrothermal degradation. The main reaction products were analytically discovered from experimentally obtained product phases. The focus of this work laid on the discovery of the main reaction pathways and the identification of the bottlenecks. Employing the tool of modelling of the formal kinetics of the discovered reactions, the investigation of reaction mechanisms was possible. In addition the influences of the application of different solvents, that are water and ethanol, and different catalysts were investigated. Finally, the economic feasibility of different process scenarios was studied in order to evaluate the attractiveness of a process which aims at the production of phenol from lignin.

Thermodynamic studies of the lignin degradation showed that even at elevated residence times of 18 h in a batch reactor a system comprising lignin and either water or ethanol had not reached thermodynamic equilibrium. Hence, a kinetic modelling of the degradation was decided to be more convenient to model the yields of major products at reasonable residence times. The simulation of the equilibrium with Aspen Plus V7.2 showed that the major part of the organic input material including 100 % of the solvent ethanol at 633 K would be converted into gaseous components. The gasiﬁcation of ethanol is thus problematic for the solvolysis of lignin in ethanol. Lignin depolymerisation is faster in water than in ethanol at moderate temperatures (633 K). This was proven by the evaluation of FT-IR-analysis of the solid residues at diﬀerent residence times. However, the reaction rate of lignin degradation does not exclusively depend on the solvent and the temperature. The type of lignin, the separation method, the thermal degradation

111 method and the heating rate also play an important role [Cho+12; Var+97; JNB10]. Furthermore, the comparison of both solvents, water and ethanol, showed that water leads to a narrower spectrum of phenolic products and a tenfold increase of the catechol yield. This is advantageous in respect to the separation of the phenolic product from solvent and by-products. The gasification of ethanol at temperatures above 633 K is significant at residence times above 120 min in a batch reactor. The gas yield increases to approximately 70 wt% at 8 h residence time on an initial lignin basis. Experimental results from a CSTR showed a threefold increase of C2-gases, that are ethane and ethene, to over 10 wt% on an initial lignin basis, if the temperature was increased from 633 K to 400 K at less than 1 h mean residence time. These facts show that the gasification of ethanol is critical at temperatures above 633 K and has to be considered for the design of a technical application.

Furthermore, the effects of heterogeneous metal catalysts on the hydrothermal lignin depolymerisation were studied. Heterogeneous catalysts have minor influence on the primary lignin depolymerisation. However, the effects on the consecutive reactions, such as the hydrodeoxygenation of phenolic intermediates are significant. For example, PdCl2 catalyses the conversion of methoxyphenols into catechols. In respect to the hydrothermal lignin degradation, the application of Raney-Ni seems especially interesting since it selectively catalyses the conversion of catechol into phenol since among the phenolic products recovered from the lignin degradation phenol finds currently the most applications in industry. Hence, the kinetics of the formation and decomposition of phenol within the hydrothermal decomposition of catechol catalysed by Raney-Ni was studied in more detail. It was found that the presence of Raney-Ni significantly increases the conversion of catechol to 88 % and the selectivity of phenol to 90 % at moderate temperatures of approximately 523 K. For comparison, the non-catalysed conversion of catechol into phenol yields a maximum selectivity of 34 % at 15 % conversion at 683 K [WSG09]. The catalysis of primary lignin degradation by cleavage of the ether bonds was not observed employing heterogeneous catalysts. However, both, basic and acidic homogeneous catalysts have a high potential to influence the mentioned reaction [Beh+06]. First screening experiments were successful and showed a threefold increase of the yield of catechols in the presence of KOH and NaOH (see Appendix D.1). HCl had an influence also on the yield of methoxyphenols as well as CO and CO2. In order to intensify the knowledge in this sector, further investigation effort is recommended.

The focus of this work laid on the development of a model containing the

112 Chapter 8. Conclusion main reaction pathways of the lignin depolymerisation. The formal kinetics defined by this model should be able to describe the phenomena which were observed within the experimental part of this work, especially the product spectrum of phenolic products, but also considering bulk products such as char or gas. The modelling was carried out for thermal degradation of wheat- straw-lignin in ethanol and formic acid as well as for the thermal degradation of spruce-lignin in water. Assumptions based on the study of literature and the observation of experimental results, yielded a series of straightforward reaction pathways. The evaluation of the optimised formal kinetic parameters of the previously defined reactions showed a very good correspondence with values from literature, especially for the hydrodeoxygenation of phenolic intermediates such as methoxyphenols and catechols. In addition, the model is able to describe the yields of phenolic products and the development of the yields over reaction time. The applicability of the ethanol model on both, batch reactor and CSTR, was proven, although the formal kinetic parameters had to be slightly modified. The model was able to describe the basic tendencies of phenolic yields. The resulting apparent kinetic parameters of the formation and decomposition of phenolics agree with values from literature. The modelling reveals the basic differences between the two different solvents, the two different types of lignin and the effect of two different reactors. The degradation of wheat-straw-lignin yields relatively high amounts of the primary product 4-ethylphenol. This is due to the structure of wheat- straw-lignin which comprises an elevated percentage of para-coumaryl units. 4-Ethylphenol is derived from para-coumaryl alcohol. Ethylation of phenol occurs at higher residence times and contributes less to the formation of ethylphenols. Consequently, the most prominent primary phenolic products from the degradation of spruce-lignin are methoxyphenols since it mainly comprises coniferyl units. The build-up of char is suppressed if the lignin degradation is transferred from a batch-reactor to a CSTR. The coexistence of early and late products of thermal lignin degradation in a CSTR facilitates the scavenging of reactive substances, e.g. radicals. This might also be the reason for decreasing gas yields and increasing yields of phenolic products from 5 to over 6 wt% on an initial lignin basis. The repolymerisation of phenolic products, basically catechols, were of minor importance according the model of hydrothermal degradation of spruce-lignin. The formation of char was more likely caused by the combination of reactive lignin-derived intermediates. The modelling shows the bottleneck reactions of thermal lignin degradation which inhibit the formation of larger amounts of phenolics. Especially the primary formation of methoxyphenols is much slower than the degradation into other reactive fragments which further react into char or gas. However, modelling also revealed that a secondary pathway

113 leads from these lignin derived fragments to the phenolic products. The latter is favoured in a CSTR. Furthermore, it could also be proven that the mechanism of the cleavage of the C-O bonds present in the methoxyl-group is facilitated in water at moderate temperatures of 633 K. Hydrolysis is the dominant mechanism in the presence of water. However, the mechanism changes around the critical point of water (Tc = 646 K). In SCW thus the cleavage of C-O bonds is radical induced. The modelling of the formal kinetics of thermal lignin degradation is a powerful tool which aids in understanding the reaction bottlenecks and the reaction mechanisms. The models introduced here, however, have to be further refined and amplified in order to describe e.g. the composition of the gaseous product and to elucidate the reactions which lead to the increased yield of catechols in water. This requires additional experimental studies. The tools for evaluation and modelling can be find in this work.

The production of phenol from lignin by solvolysis in ethanol or hydrothermal degradation is economically not feasible if the current yields and process parameters as well as the current market prices for the potential products are considered. The most important parameter inﬂuencing the feasibility is the phenol yield. However, even if the phenol yield could be increased to 30 wt% referring to the input amount of lignin, which seems to be the maximum at the current state of the art, a process would still not be economically attractive. A reduction of the mean residence time in the reactor and an increase of the lignin/solvent ratio in the feed are also key parameters which contribute to the necessary drastic reduction of the capital related costs. The yield of phenol and other phenolics has to be increased to at least 20 %, the lignin/solvent ratio should be increased signiﬁcantly (20 to 30 %) and the mean residence time has to be reduced to maximum 10 to 30 min in order to make the process economically attractive. However, the feasibility study provides a rough and rather conservative estimation of the capital related costs. Hydrothermal lignin degradation shows better economic opportunities than solvolysis in ethanol since the feedstock situation provides large quantities of cheap feedstock, the solvent is more stable and abundantly available at low prices. Future research should thus focus on black liquor. However, the yield of phenol from black liquor will be reduced since it usually contains more than 50 wt% residues, e.g. cellulose and hemicellulose.

The present work shows that hydrothermal depolymerisation of lignin has a certain potential for the recovery of phenolic platform chemicals. It has signiﬁcant advantages over thermal lignin depolymerisation in organic

114 Chapter 8. Conclusion solvents, e.g. ethanol. Furthermore the potential of catalysts in combination with water is considerable. A combination of a homogeneous catalyst favour- ing the cleavage of C-O bonds and a heterogeneous catalysts for the adjacent valorisation of phenolic products is a promising approach to maximise the yield of valuable phenolics. However, at the current state of research a technical process is not recommended since it is economically not attractive.

Future research eﬀort in respect to the hydrothermal lignin degradation should focus on

• black liquor since it is the most abundantly available waste product containing lignin in an aqueous solution. Furthermore, it provides the possibility to realise a grave to cradle process since the feedstock is a waste material. • A more intensive investigation of the influence of heterogeneous and homogeneous catalysis is recommended in order to maximise the yield of phenolics. • In respect to the study of kinetics, the model developed in the present work should be amplified in order to provide a tool covering different feedstocks and a detailed prediction of the formation of gaseous by-products and solid residue. Here the focus should be the lump- component LD since its constituents play a central role for the lignin degradation.

115 116 Bibliography

[ADL10] P. Al-Riﬀai, B. Dimaranan, and D. Laborde. Global Trade and Environmental Impact Study of the EU Biofuels Mandate. Fi- nal Draft Report. International Food Policy Institute (IFPRI), Mar. 2010. [Aid+02] T. M. Aida, T. Sato, G. Sekiguchi, T. Adschiri, and K. Arai. “Extraction of Taiheiyo coal with supercritical water-phenol mixtures”. Fuel 81.11-12 (2002), pp. 1453–1461. [APR01] C. Amen-Chen, H. Pakdel, and C. Roy. “Production of monomeric phenols by thermochemical conversion of biomass: a review”. Bioresource Technology 79.3 (2001), pp. 277–299. [Arp07] H.-J. Arpe. Industrielle Organische Chemie: Bedeutende Vor- und Zwischenprodukte. 6th ed. Wiley-VCH Verlag GmbH & Co. KGaA, 2007. isbn: 9783527315406. [Asp10] Aspen. Aspen Physical Property System Methods. Aspen Tech- nology, Inc. 200 Wheeler Road Burlington, MA 01803-5501 USA, 2010. [AST87] ASTM. “Standard Speciﬁcation for Aviation Turbine Fuels”. Annual Book of ASTM Standards 1655.82 (Jan. 1987), pp. 835– 844. [Bae+06] M. Baerns, A. Behr, J. Gmehling, H. Hofmann, U. Onken, and A. Renken. Lehrbuch Technische Chemie. 1st ed. Wiley-VCH, 2006. isbn: 3527310002. [Baz+07] A. Bazaev, I. Abdulagatov, E. Bazaev, and A. Abdurashidova. “PVT Measurements for Pure Ethanol in the Near-Critical and Supercritical Regions”. International Journal of Thermo- physics 28.1 (2007), pp. 194–219.

117 BIBLIOGRAPHY

[Beh+06] F. Behrendt, Y. Neubauer, K. Schulz-Tönnies, B. Wilmes, and N. Zobel. Direktverflüssigung von Biomasse - Reaktionsmech- anismen und Produktverteilungen. Tech. rep. Bundesanstalt für Landwirtschaft und Ernährung and Technische Universität Berlin, June 2006. [Bei+08] N. Beintema, D. Bossio, F. Dreyfus, M. Fernandez, A. Gurib- Fakim, H. Hurni, A.-M. Izac, J. Jiggins, G. K.-. Berisavljevic, R. Leakey, W. Ochola, B. Osman-Elasha, C. Plencovich, N. Röling, M. Rosegrant, E. Rosenthal, and L. Smith. Agricul- ture at a Crossroads - Summary for Decision Makers of the Global Report. Tech. rep. International Assessment of Agri- cultural Knowledge, Science and Technology for Development (IAASTD), Apr. 2008. [BK08] T. Barth and M. Kleinert. “Motor Fuels From Biomass Pyroly- sis”. Chemical Engineering & Technology 31.5 (2008), pp. 773– 781. [BKB90] N. Bhore, M. Klein, and K. Bischoff. “The Delplot Technique: A New Method for Reaction Pathway Analysis”. Industrial & Engineering Chemistry Research 29.2 (1990), pp. 313–316. [BV10] M. Brebu and C. Vasile. “Thermal degradation of lignin - A review”. Cellulose Chemistry & Technology 44.9 (2010), pp. 353– 363. [CF11] D. Castello and L. Fiori. “Supercritical water gasification of biomass: Thermodynamic constraints”. Bioresource Technology 102.16 (2011), pp. 7574–7582. [Cho+12] J. Cho, S. Chu, P. J. Dauenhauer, and G. W. Huber. “Ki- netics and reaction chemistry for slow pyrolysis of enzymatic hydrolysis lignin and organosolv extracted lignin derived from maplewood”. Green Chemistry 14.2 (2012), pp. 428–439. [CKL09] S. Consonni, R. E. Katofsky, and E. D. Larson. “A gasification- based biorefinery for the pulp and paper industry”. Chemical Engineering Research and Design 87.9 (2009). Special Issue on Biorefinery Integration, pp. 1293 –1317. [Cos+9] L. da Costa Sousa, S. P. Chundawat, V. Balan, and B. E. Dale. “’Cradle-to-grave’ assessment of existing lignocellulose pretreatment technologies”. Current Opinion in Biotechnology 20.3 (2009), 339–347.

118 BIBLIOGRAPHY

[DIN84] DIN. Mineralöle und Brennstoﬀe 1. Grundnormen, Normen über Eigenschaften und Anforderungen. Ed. by DIN Deutsches Institut für Normung e. V. 9th ed. Beuth Verlag GmbH, 1984. isbn: 3410114599. [DM99] E. Dorrestijn and P. Mulder. “The radical-induced decomposition of 2-methoxyphenol”. Journal of the Chemical Society. Perkin Transactions 2.4 (1999), pp. 777–780. [Dor+00] E. Dorrestijn, L. J. Laarhoven, I. W. Arends, and P. Mulder. “The occurrence and reactivity of phenoxyl linkages in lignin and low rank coal”. Journal of Analytical and Applied Pyrolysis 54.1-2 (2000), pp. 153–192. [Dor+99] E. Dorrestijn, M. Kranenburg, D. Poinsot, and P. Mulder. “Lignin depolymerization in hydrogen-donor solvents”. Holzforschung 53 (1999), pp. 611–616. [DP04] H. Dillon and S. Penoncello. “A Fundamental Equation for Cal- culation of the Thermodynamic Properties of Ethanol”. Inter- national Journal of Thermophysics 25.2 (2004), pp. 321–335. [DRS74] D. Domburg, G. Rossinakaya, and V. Sergeeva. “Study of thermal stability of b-ether bonds in lignin and its models”. 4th International Conference on Thermal Analysis. 1974, pp. 211– 221. [EEA06] EEA. How much bioenergy can Europe produce without harming the environment? Tech. rep. European Environment Agency (EEA), July 2006. [Ehr12] C. Ehrenstein. “Wenn der Acker nur noch Strom erzeugt”. Die Welt 3 (Jan. 2012). [EIA11] EIA. Coal Production and Preparation Report. Tech. rep. U.S. Energy Information Administration, Nov. 2011. [Fan+05] Z. Fang, T. Sato, R. Smith, H. Inomata, K. Arai, and J. Kozin- ski. “Reaction chemistry and phase behavior of lignin in high- temperature and supercritical water”. Bioresource Technology 99 (2005), pp. 3424–3430. [Far+10] T. Faravelli, A. Frassoldati, G. Migliavacca, and E. Ranzi. “De- tailed kinetic modeling of the thermal degradation of lignins”. Biomass and Bioenergy 34.3 (2010), pp. 290–301. [FJP75] A. Fredenslund, R. L. Jones, and J. M. Prausnitz. “Group- contribution estimation of activity coeﬃcients in nonideal liquid mixtures”. AIChE Journal 21.6 (1975), pp. 1086–1099.

119 BIBLIOGRAPHY

[FMG87] O. Faix, D. Meier, and I. Grobe. “Studies on isolated lignins and lignins in woody materials by Pyrolysis - Gas Chromatography - Mass Spectrometry and off-line Pyrolysis - Gas Chromatog- raphy with Flame Ionization Detection”. Journal of Analytical and Applied Pyrolysis 11 (1987), pp. 403–416. [FN68] K. Freudenberg and A. Neish. Constitution and Biosynthesis of Lignin. 2nd ed. Springer Verlag, 1968. isbn: 9783540042747. [FNR11] FNR. Maisanbau in Deutschland. Tech. rep. Fachagentur Nachwach- sende Rohstoffe e.V. (FNR), Apr. 2011. [For+12] D. Forchheim, U. Hornung, P. Kempe, A. Kruse, and D. Stein- bach. “Influence of RANEY Nickel on the formation of intermediates in the degradation of lignin”. International Journal of Chemical Engineering (2012), Article ID 589749. [FW87] D. Fengel and G. Wegener. Wood Chemistry Ultrastructure Re- actions. 1st ed. Verlag Kessel, 1987. isbn: 9783935638395. [Gas+10] J. Gasson, M. Kleinert, T. Barth, D. Forchheim, E. Sahin, A. Kruse, and I. Eide. “Lignin Solvolysis: Upscaling of the Lignin- to-Liquid Conversion Process towards Technical Applicability”. Proceedings of the 18th European Biomass Conference. 2010, pp. 10–13. [Gas+2] J. R. Gasson, D. Forchheim, T. Sutter, U. Hornung, A. Kruse, and T. Barth. “Modeling the Lignin Degradation Kinetics in an Ethanol/Formic Acid Solvolysis Approach. Part 1. Kinetic Model Development”. Industrial & Engineering Chemistry Re- search 51.32 (Aug. 2012), 10595–10606. [Gel+08] G. Gellerstedt, J. Li, I. Eide, M. Kleinert, and T. Barth. “Chem- ical Structures Present in Biofuel Obtained from Lignin”. En- ergy & Fuels 22.6 (2008), pp. 4240–4244. [Gla09] W. Glasser. “Lignin - Retrospect and Prospect”. Gülzower Fachge- spräche 31.7 (2009), pp. 42–44. [Gme+02] J. Gmehling, R. Wittig, J. Lohmann, and R. Joh. “A Modified UNIFAC (Dortmund) Model. 4. Revision and Extension”. In- dustrial & Engineering Chemistry Research 41.6 (2002), pp. 1678– 1688. [Gut+09] A. Gutierrez, R. Kaila, M. Honkela, R. Slioor, and A. Krause. “Hydrodeoxygenation of guaiacol on noble metal catalysts”. Catalysis Today 147.3-4 (2009), pp. 239–246.

120 BIBLIOGRAPHY

[Haa2] M. Haase. “Entwicklung eines Energie- und Stoffstrommodells zur ökonomischen und ökologischen Bewertung der Herstellung chemischer Grundstoffe aus Lignocellulose”. PhD thesis. Karl- sruhe Institute of Technology, 2012. [Her71] H. L. Hergert. “Infrared Spectra”. Lignins: Occurrence, Forma- tion, Structure and Reaction. Ed. by K. Sarkanen and C. Lud- wig. 1st ed. Wiley Interscience, New York, USA, 1971, pp. 268– 272. [HKB12] B. Holmelid, M. Kleinert, and T. Barth. “Reactivity and Reac- tion Pathways in Thermochemical Treatment of Selected Lignin Model Compounds under Hydrogen Rich Conditions”. Journal of Analytical & Applied Pyrolysis 98 (Nov. 2012), pp. 37–44. [HS2] C. M. Huelsman and P. E. Savage. “Intermediates and kinetics for phenol gasification in supercritical water”. Physical Chem- istry Chemical Physics 14.8 (2012), pp. 2900–2910. [Hum+1] D. Humbird, R. Davis, L. T. C. Kinchin, D. Hsu, A. Aden, P. Schoen, J. Lukas, B. Olthof, M. Worley, D. Sexton, and D. Dudgeon. Process Design and Economics for Biochemical Conversion of Lignocellulosic Biomass to Ethanol: Dilute-Acid Pretreatment and Enzymatic Hydrolysis of Corn Stover. Tech. rep. National Renewable Energy Laboratory (NREL), Golden, CO., 2011. [Imh12] Imhoff. Oral offer from IMHOFF & STAHL GmbH. Nov. 2012. [JK85] H. E. Jegers and M. T. Klein. “Primary and secondary lignin pyrolysis reaction pathways”. Industrial & Engineering Chem- istry Process Design and Development 24 (1985), pp. 173–183. [JNB10] G. Jiang, D. J. Nowakowski, and A. V. Bridgwater. “A systematic study of the kinetics of lignin pyrolysis”. Thermochimica Acta 498 (2010), pp. 61–66. [Kar+06] S. Karagöz, T. Bhaskar, A. Muto, and Y. Sakata. “Hydrother- mal upgrading of biomass: Effect of K2CO3 concentration and biomass/water ratio on products distribution”. Bioresource Tech- nology 97.1 (Jan. 2006), pp. 90–98. [KB08a] M. Kleinert and T. Barth. “Phenols from Lignin”. Chemical Engineering & Technology 31.5 (2008), pp. 736 –745. [KB08b] M. Kleinert and T. Barth. “Towards a Lignincellulosic Biorefin- ery: Direct One-Step Conversion of Lignin to Hydrogen-Enriched Biofuel”. Energy Fuels 22 (2008), pp. 1371–1379.

121 BIBLIOGRAPHY

[KGB09] M. Kleinert, J. R. Gasson, and T. Barth. “Optimizing solvolysis conditions for integrated depolymerisation and hydrodeoxygenation of lignin to produce liquid biofuel”. Journal of Ana- lytical and Applied Pyrolysis 85.1-2 (2009), pp. 108–117. [KHB12] M. Klein, Z. Hou, and C. Bennett. “Reaction Network Eluci- dation: Interpreting Delplots for Mixed Generation Products”. Energy Fuels 26.1 (2012), pp. 52–54. [Kle+11] M. Kleinert, J. Gasson, I. Eide, A.-M. Hilmen, and T. Barth. “Developing Solvolytic Conversion of Lignin-to-Liquid (LtL) Fuel Components: Optimization of Quality and Process Fac- tors”. Cellulose Chemistry & Technology 45 (2011), pp. 3–12. [Kog09] K. Kogure. Lead Storage Battery. Patent JP2009129725. 2009. [Kru+00] A. Kruse, D. Meier, P. Rimbrecht, and M. Schacht. “Gasiﬁ- cation of Pyrocatechol in Supercritical Water in the Presence of Potassium Hydroxide”. Industrial & Engineering Chemistry Research 39.12 (2000), pp. 4842–4848. [KT89] A. Klemola and J. Touvinen. Method for the production of vanillin. Patent US4847422. 1989. [KV08] M. Klein and P. Virk. “Modeling of Lignin Thermolysis”. En- ergy Fuels 22 (2008), pp. 2175–2182. [KV80] M. Klein and P. Virk. “Model Pathways for Gas Release from Lignites”. Preprints of Papers - American Chemical Society, Division of Fuel Chemistry 25.4 (1980), pp. 180–190. [Lag+98] J. Lagarias, J. Reeds, M. Wright, and P. Wright. “Convergence Properties of the Nelder–Mead Simplex Method in Low Dimen- sions”. SIAM Journal on Optimization 9.1 (1998), pp. 112–147. [Lev99] O. Levenspiel. Chemical Reaction Engineering. 3rd ed. Wiley- VCH, 1999. isbn: 9780471254249. [LI92] S. Lau and R. Ibrahim. “FT-IR Spectroscopic Studies on Lignin from Some Tropical Woods and Rattan”. Pertanika 14.1 (1992), pp. 75–81. [LK85] J. Lawson and M. Klein. “Inﬂuence of Water on Guaiacol Py- rolysis”. Industrial and Engineering Chemistry, Fundamentals 24 (1985), pp. 203–208. [LL1] K. Lundquist and R. Lundgren. “Acid Degradation of Lignin. Part VII. The Cleavage of Ether Bonds.” Acta Chemica Scan- dinavica 26 (1971), 2005–2023.

122 BIBLIOGRAPHY

[LP43] W. Lautsch and G. Piazolo. “Über die Hydrierung von Lignin und ligninhaltigen Stoffen mit Wasserstoff abgebenden Mitteln, insbesondere Alkoholen”. Berichte der deutschen chemischen Gesellschaft (A and B Series) 76.5 (1943), pp. 486–498. [LPH97] C. E. López Pasquali and H. Herrera. “Pyrolysis of Lignin and IR Analysis of Residues”. Thermochimica Acta 293 (1997), pp. 39–46. [Lun6] K. Lundquist. “Low-molecular weight lignin hydrolysis products”. Applied Polymer Symposium 28 (1976), 1393–1407. [LWC11] K. Li, R. Wang, and J. Chen. “Hydrodeoxygenation of Anisole over Silica-Supported Ni2P, MoP, and NiMoP Catalysts”. En- ergy & Fuels 25.3 (2011), pp. 854–863. [LZ08] Z. Liu and S. Zhang. “Effects of various solvents on the liquefaction of biomass to produce fuels and chemical feedstocks”. Energy Conversion and Management 49.12 (2008), pp. 3498 – 3504. [MB97] R. Miller and J. Bellan. “A Generalized Biomass Pyrolysis Model Based on Superimposed Cellulose, Hemicellulose and Lignin Kinetics”. Combustion Science & Technology 126 (1997), pp. 97–137. [McM+04] D. F. McMillen, R. Malhotra, S.-J. Chang, S. E. Nigenda, and G. A. S. John. “Coupling pathways for dihydroxy aromatics during coal pyrolysis and liquefaction”. Fuel 83.11-12 (2004), pp. 1455–1467. [Mil+99] J. Miller, L. Evans, A. Littlewolf, and D. Trudell. “Batch Mi- croreactor Studies of Lignin and Lignin Model Compound De- polymerization by Bases in Alcohol Solvents”. Fuel 78 (1999), pp. 1363–1366. [Moh06] D. Mohan. “Pyrolysis of Wood/Biomass for Bio-Oil: A Critical Review”. Energy Fuels 20 (2006), pp. 848–889. [MS1] D. Meier and W. Schweers. “Properties and decomposition of lignins isolated by means of an alcoholic water-mixtures .4. Production of monomeric phenols by catalytic hydrogenolysis”. Holzforschung 35.2 (1981), 81–85. [Nim+11a] T. Nimmanwudipong, R. Runnebaum, D. Block, and B. Gates. “Catalytic Conversion of Guaiacol Catalyzed by Platinum Sup- ported on Alumina: Reaction Network Including Hydrodeoxy- genation Reactions”. Energy Fuels 25 (2011), pp. 3417–3427.

123 BIBLIOGRAPHY

[Nim+11b] T. Nimmanwudipong, R. Runnebaum, D. Block, and B. Gates. “Catalytic Reactions of Guaiacol: Reaction Network and Evi- dence of Oxygen Removal in Reactions with Hydrogen”. Catal- ysis Letters 141.6 (2011), pp. 779–783. [Now+10] D. Nowakowski, A. Bridgwater, D. Elliott, D. Meier, and P. de Wild. “Lignin fast pyrolysis: Results from an international collaboration”. Journal of Analytical and Applied Pyrolysis 88.1 (2010), pp. 53 –72. [Nun+85] T. Nunn, J. Howard, J. Longwell, and W. Peters. “Product Compositions and Kinetics in the Rapid Pyrolysis of Milled Wood Lignin”. Industrial & Engineering Chemistry Process De- sign and Development 24 (1985), pp. 844–852. [OR11] S. J. Okullo and F. Reynès. “Can reserve additions in mature crude oil provinces attenuate peak oil?” Energy 36.9 (2011), pp. 5755 –5764. [Pea67] I. Pearl. The Chemistry of Lignin. 1st ed. Marcel Dekker Inc., New York, USA, 1967. isbn: 0824715454. [Pfe04] A. Pfennig. “GE-Modelle”. Thermodynamik der Gemische. Ed. by A. Pfennig. 1st ed. Springer, 2004. Chap. 5. [PK08] N. Prakash and T. Karunanithi. “Kinetic Modeling in Biomass Pyrolysis - A Review”. Journal of Applied Sciences Research 4.12 (2008), pp. 1627–1636. [PK84] F. Petrocelli and M. Klein. “Model Reaction Pathways in Kraft Lignin Pyrolysis”. Macromolecules 17 (1984), pp. 161–169. [PR76] D.-Y. Peng and D. B. Robinson. “A New Two-Constant Equa- tion of State”. Industrial Engineering Chemistry Fundamentals 15.1 (1976), pp. 59 –63. [PT04] M. Peters and K. Timmerhaus. Plant Design and Economics for Chemical Engineers. Ed. by E. D. Glandt, M. T. Klein, and T. F. Edgar. 5th ed. McGraw-Hill, 2004. isbn: 0072392665. [Pul09] J. Puls. “Lignin - Verfügbarkeit, Markt und Verwendung: Per- spektiven für schwefelfreie Lignine”. Gülzower Fachgespräche 31.5 (2009), pp. 18–41. [Ram70] M. Ramiah. “Thermogravimetric and diﬀerential thermal analysis of cellulose, hemicellulose, and lignin”. Journal of Applied Polymer Science 14 (1970), pp. 1323–1337.

124 BIBLIOGRAPHY

[RBS87] A. Roy, S. Bag, and S. Sen. “Studies on the Chemical Nature of Milled Wood Lignin of Jute Stick”. Cellulose Chemistry & Technology 21 (1987), pp. 343–348. [Rob+1] V. M. Roberts, V. Stein, T. Reiner, A. Lemonidou, X. Li, and J. A. Lercher. “Towards Quantitative Catalytic Lignin Depoly- merization”. Chemistry-A European Journal 17.21 (May 2011), 5939–5948. [RS09] R. Rinaldi and F. Schuth. “Design of solid catalysts for the conversion of biomass”. Energy & Environmental Science 2.6 (2009), pp. 610–626. [Run+12] R. C. Runnebaum, T. Nimmanwudipong, D. E. Block, and B. C. Gates. “Catalytic conversion of compounds representa- tive of lignin-derived bio-oils: a reaction network for guaiacol, anisole, 4-methylanisole, and cyclohexanone conversion catalysed by Pt/γ-Al2O3”. Catalysis Science & Technology 2 (2012), pp. 113–118. [RW12] R. Rinaldi and X. Wang. “Hydrogenolysis of Lignin to What You Want - The Role of Solvent”. 5th TMFB International Workshop: Tailor-Made Fuels from Biomass. Max-Planc-Insitut für Kohlenforschung. 2012. [Sai+03] M. Saisu, T. Sato, M. Watanabe, T. Adschiri, and K. Arai. “Conversion of Lignin with Supercritical Water-Phenol Mix- tures”. Energy & Fuels 17.4 (2003), pp. 922–928. [Sch07] K. Schwarz. Lignin-based Glue. Patent WO2007048494. 2007. [Sch12] C. Schröder. “Volvo Trucks: Erste Zwischenergebnisse beim Bio-DME-Projekt übertreﬀen die Erwartungen”. Automobiltech- nische Zeitschrift (Mar. 2012). [SH11] A. G. Sergeev and J. F. Hartwig. “Selective, Nickel-Catalyzed Hydrogenolysis of Aryl Ethers”. Science 6028.332 (Apr. 2011), pp. 439–443. [SKR04] A. Sınaˇg,A. Kruse, and J. Rathert. “Inﬂuence of the Heating Rate and the Type of Catalyst on the Formation of Key Inter- mediates and on the Generation of Gases During Hydropyrol- ysis of Glucose in Supercritical Water in a Batch Reactor”. In- dustrial & Engineering Chemistry Research 43.2 (2004), pp. 502– 508.

125 BIBLIOGRAPHY

[SR11] W. Stinner and N. Rensberg. Perspektive der Biogasgewinnung aus nachwachsenden Rohstoffen - Bewertung von Substratalter- nativen. Study. Deutsches BiomasseForschungsZenrum (DBFZ), Apr. 2011. [Tak+12] S. Takami, K. Okuda, X. Man, M. Umetsu, S. Ohara, and T. Adschiri. “Kinetic Study on the Selective Production of 2- (Hydroxybenzyl)-4-methylphenol from Organosolv Lignin in a Mixture of Supercritical Water and p-Cresol”. Industrial & En- gineering Chemistry Research 51.13 (2012), pp. 4804–4808. [Tal81] A. Talwalkar. Coal Conversion Systems Technical Data Book. Tech. rep. Institute of Gas Technology, Chicago, IL (USA), Nov. 1981. [Tow+07] M. Towers, T. Browne, R. Kerekes, J. Paris, and H. Tran. “Biorefinery opportunities for the Canadian pulp and paper industry”. Pulp Paper Canada 6.6 (2007), pp. 109–112. [TRR11] S. S. Toor, L. Rosendahl, and A. Rudolf. “Hydrothermal liquefaction of biomass: A review of subcritical water technologies”. Energy 36.5 (2011), pp. 2328–2342. [Var+97] G. Várhegyi, M. Antal, E. Jakab, and P. Szabó. “Kinetic Mod- eling of Biomass Pyrolysis”. Journal of Analytical and Applied Pyrolysis 42.1 (1997), pp. 73–87. [VCI11] VCI. Chemiewirtschaft in Zahlen. Tech. rep. Verband der Chemis- chen Industrie (VCI), 2011. [VCI12] VCI. Preisindex - Chemie Technik. 2012. [VDI06] VDI. VDI-Wärmeatlas. Ed. by Verein Deutscher Ingenieure (VDI) - Gesellschaft Verfahrenstechnik und Chemieingenieur- wesen (GVC). Springer-Verlag, 2006. isbn: 3540255044. [Vuo86] A. Vuori. “Pyrolysis Studies of some Simple Coal Related Aro- matic Methyl Esters”. Fuel 65 (1986), pp. 1575–1583. [Wag09] W. Wagner. Wärmeaustauscher: Grundlagen, Aufbau und Funk- tion thermischer Apparate. 4th ed. Vogel Business Media, 2009. isbn: 3834331619. [WGZ03] A. Wingren, M. Galbe, and G. Zacchi. “Techno-Economic Eval- uation of Producing Ethanol from Softwood: Comparison of SSF and SHF and Identification of Bottlenecks”. Biotechnology Progress 19.4 (2003), pp. 1109–1117.

126 BIBLIOGRAPHY

[Wil+09] P. de Wild, R. Van der Laan, A. Kloekhorst, and E. Heeres. “Lignin valorisation for chemicals and (transportation) fuels via (catalytic) pyrolysis and hydrodeoxygenation”. Environmental Progress & Sustainable Energy 28.3 (2009), pp. 461–469. [WR12] X. Wang and R. Rinaldi. “Solvent Eﬀects on the Hydrogenolysis of Diphenyl Ether with Raney Nickel and their Implications for the Conversion of Lignin”. ChemSusChem 5.8 (2012), pp. 1455– 1466. [WSG08] Wahyudiono, M. Sasaki, and M. Goto. “Recovery of phenolic compounds through the decomposition of lignin in near and supercritical water”. Chemical Engineering and Processing: Pro- cess Intensiﬁcation 47.9-10 (2008), pp. 1609–1619. [WSG09] Wahyudiono, M. Sasaki, and M. Goto. “Conversion of biomass model compound under hydrothermal conditions using batch reactor”. Fuel 88.9 (2009), pp. 1656–1664. [WSG11] Wahyudiono, M. Sasaki, and M. Goto. “Thermal Decomposi- tion of Guaiacol in Sub- and Supercritical Water and its Kinetic Analysis”. Journal of Material Cycles and Waste Management 13 (2011), pp. 68–79. [Ye+12] Y. Ye, Y. Zhang, J. Fan, and J. Chang. “Novel Method for Production of Phenolics by Combining Lignin Extraction with Lignin Depolymerization in Aqueous Ethanol”. Industrial & Engineering Chemistry Research 51.1 (2012), pp. 103–110. [YM12] T. L.-K. Yong and Y. Matsumura. “Reaction Kinetics of the Lignin Conversion in Supercritical Water”. Industrial & Engi- neering Chemistry Research 51.37 (2012), pp. 11975–11988. [Yok+98] C. Yokoyama, K. Nishi, A. Nakajima, and K. Seino. “Thermol- ysis of organosolv lignin in supercritical water and supercritical methanol”. Sekiyu Gakkai 41.4 (1998), pp. 243–250. [YS98] J. Yu and P. Savage. “Decomposition of Formic Acid under Hy- drothermal Conditions”. Industrial & Engineering Chemistry Research 37.1 (1998), pp. 2–10. [Yua+10] Z. Yuan, S. Cheng, M. Leitch, and C. Xu. “Hydrolytic Degrada- tion of Alkaline Lignin in Hot-Compressed Water and Ethanol”. Bioresource Technology 101.23 (2010), pp. 9308–9313.

127 BIBLIOGRAPHY

[Zak+0] J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius, and B. M. Weckhuysen. “The Catalytic Valorization of Lignin for the Pro- duction of Renewable Chemicals”. Chemical Reviews 110.6 (June 2010), 3552–3599. [Zha+1] C. Zhao, J. He, A. A. Lemonidou, X. Li, and J. A. Lercher. “Aqueous-phase hydrodeoxygenation of bio-derived phenols to cycloalkanes”. Journal of Catalysis 280.1 (May 2011), 8–16. [Zha+10a] C. Zhao, Y. Kou, A. A. Lemonidou, X. Li, and J. A. Lercher. “Hydrodeoxygenation of bio-derived phenols to hydrocarbons using RANEY-Ni and Naﬁon/SiO2 catalysts”. Chemical Com- munications 46 (2010), pp. 412–414. [Zha+10b] W. Zhao, W.-J. Xu, X.-J. Lu, C. Sheng, S.-T. Zhong, S.-R. Tang, Z.-M. Zong, and X.-Y. Wei. “Preparation and Property Measurement of Liquid Fuel from Supercritical Ethanolysis of Wheat Stalk”. Energy & Fuels 24.1 (2010), pp. 136–144. [Zha+11] H. Zhao, D. Li, P. Bui, and S. Oyama. “Hydrodeoxygenation of guaiacol as model compound for pyrolysis oil on transition metal phosphide hydroprocessing catalysts”. Applied Catalysis A: General 391.1-2 (2011), pp. 305–310. [ZHR11] B. Zhang, H.-J. Huang, and S. Ramaswamy. “Reaction Kinetics of the Hydrothermal Treatment of Lignin”. Applied Biochem- istry & Biotechnology 147.1-3 (2011), pp. 119–131. [ZL12] C. Zhao and J. A. Lercher. “Upgrading Pyrolysis Oil over Ni/HZSM- 5 by Cascade Reactions”. Angewandte Chemie 124.24 (2012), pp. 6037–6042. [ZL2] C. Zhao and J. A. Lercher. “Selective Hydrodeoxygenation of Lignin-Derived Phenolic Monomers and Dimers to Cycloalka- nes on Pd/C and HZSM-5 Catalysts”. ChemCatChem 4.1 (Jan. 2012), 64–68.

128 List of Tables

1.1 BDE of the most predominant linkages (see Figure 1.1.)...5

2.1 Characterisation of both lignin types from wheat straw and spruce...... 14

3.1 Initial yields yi,0 for the Aspen-calculation of thermodynamic equilibrium at 633 K, derived from the experimental results at 120 min in water (Exp. 90) and ethanol (Exp. 158, see Appendix E.1)...... 23 3.2 Equilibrium yields yi of thermal lignin degradation in water and ethanol calculated with Unifac and Peng-Robinson-EoS at 633 K and 30 MPa, input concentrations are listed in Ta- ble 3.1...... 24

4.1 Identiﬁed products with corresponding retention time in the GC-FID...... 33

6.1 Characteristic FT-IR absorptions and observed changes between the biomass feed and the solid residue samples..... 51 6.2 Coefficient of determination R2 for the model fit and standard deviation σ of calculated and experimental yields (see Appendix C.4)...... 57 6.3 Formal kinetic rate coefficients (kj) of all 17 reactions defined within the developed model (see Figure 6.7)...... 59 6.4 Experimental conditions, estimated total reaction mass (m) and mass flow (m˙ ) for batch and continuous reactor...... 63 6.5 Coefficient of determination R2 for the model fit and standard deviation σ of calculated and experimental yields of phenolic products (see Figure 6.12)...... 67

129 LIST OF TABLES

6.6 Formal kinetic rate coefficients kj, apparent activation energies EA,j and Arrhenius factors Aj of the 17 reactions defined in the model (see Figure 6.7) at different temperatures for the CSTR-fit are compared with rate coefficients calculated for the batch-fit at 633 K...... 68 6.7 Apparent activation energies EA,j, Arrhenius factors Aj and rate coefficients kj at different temperatures for the 15 reactions defined in the model (see Figure 6.17)...... 82 6.8 Comparison of different kinetic studies of thermal lignin degradation, all based on lump-models; values obtained from this work refer to a lump of the reactions represented by k1+k7 in the water model (Figure 6.17) and by k1+k6+k10 in the ethanol model (Figure 6.7)...... 88 6.9 Comparison of different kinetic studies of thermal decomposition of methoxyphenols, lump of the reactions represented by k2+k3 in the water-model (Figure 6.17) and by k2 in the ethanol-model (Figure 6.7), and catechols, lump of the reactions represented by k4+k5+k12 in the water model...... 91

7.1 Streams deﬁned for the lignin liquefaction process (see Fig- ure 7.1) for both, ethanol and water approach...... 96 7.2 Parameters used for the estimation of the purchased costs of the pump...... 97 7.3 Reaction conditions and resulting reactor volume VR for both ethanol and water approach...... 98 7.4 Data concerning heat exchanger (B and F) for both ethanol and water approach...... 99 7.5 Estimated prices for reactants, products and utilities...... 101 7.6 Total Capital Investment TCI calculated by various factors based on TBM...... 103 7.7 Energy demands and utility costs (TUC)...... 104 7.8 Operating Labor Costs (OLC)...... 104 7.9 Total Manufacturing Costs COM...... 106 7.10 Scenarios of the plant design and cost estimation for solvolytic lignin liquefaction and phenol production...... 109

A.1 Chemicals used...... 140 A.2 Catalysts...... 141 A.3 Construction materials...... 141

B.1 Applied values of fd,i for quantiﬁed phenolics i...... 149

130 LIST OF TABLES

B.2 Distribution of phenolic products in a two phase system comprising water and ethyl acetate under diﬀerent conditions... 150

C.1 Indices of reaction participant...... 155 C.2 Indices of reaction participant...... 157

D.1 Parameters describing the decomposition of catechol and phenol within hydrothermal catechol decomposition catalysed by Raney-Ni...... 170

E.1 Experimental data: reaction conditions, input parameters, employed analytical methods...... 175 E.2 Experimental data: Results of bulk components, gaseous compounds, H2O and EtOH...... 181 E.3 Experimental data: Results of phenolic and other compounds quantiﬁed viaGC-FID...... 187 E.4 Heating value and elemental composition of input and output components for solvolysis of wheat-straw lignin in ethanol and formic acid in a CSTR at 673 K, 25 MPa and 44 min mean residence time...... 194 E.5 Mass and energy balance for experiment Exp. 66 scaled to 1 t/h lignin...... 195

131 LIST OF TABLES

132 List of Figures

1.1 Example of lignin structure depicting some of the dominant linkages between the aromatic monomers [WR12]...... 5 1.2 The three characteristic aromatic substances which constitute the characteristic monomeric units of lignin: A: para- coumaryl alcohol, B: sinapyl alcohol and C: coniferyl alcohol.6 1.3 Monoaromatic products (phenolics) from lignin depolymerisation: A: phenol, B: catechol, C: guaiacol and D: syringol..7

3.1 Thermodynamic equilibrium calculated by Aspen with Unifac- method and Peng-Robinson-EoS compared with results from Exp. 148 a) mass fraction of of product phases and b) composition of the gas phase; τ = 1080 min; spruce-lignin/water = 132 g/L; T = 633 K...... 25 3.2 Thermodynamic equilibrium calculated by Aspen with Unifac- method and Peng-Robinson-EoS compared with results from Exp. 152 a) mass fraction of of product phases and b) composition of the gas phase; τ = 1080 min; spruce-lignin/ethanol = 132 g/L; T = 633 K...... 25

4.1 Results from thermal degradation of spruce lignin in ethanol (EtOH) or water (H2O) at T = 633 K; spruce-lignin/solvent = 132 g/L; a) yield of gas and solid residue in EtOH (solid symbols, Exp. 153-161) and H2O (open symbols, Exp. 85-93); b) Gas composition for τ = 480 min...... 30 4.2 FT-IR-analysis of spruce-lignin and the solid residue after 15, 30, 60 and 480 min residence time in a 5mLMA2 with a) ethanol (Exp. 153, 154, 156, 161) and b) H2O(Exp. 85, 86, 88, 93); lignin/solvent = 132 g/L; T = 633 K...... 31

133 LIST OF FIGURES

4.3GC-FID-chromatogram of liquid product (EtOH, Exp. 143) and organic phase after extraction of aqueous liquid product (H2O, Exp. 142) after thermal treatment of spruce-lignin; lignin/solvent = 132 g/L; FA/solvent = 107g/L; T = 633 K, τ = 60 min...... 32 4.4 Yield of diﬀerent phenolics from lignin solvolysis in ethanol (solid symbols, Exp 126-133) and H2O (open symbols, Exp 134- 141) at diﬀerent temperatures; spruce-lignin/solvent = 80 g/L; 8 mg Pd-catalyst; 1 MPa H2 initial pressure; τ = 60 min... 34

5.1 Effect of different heterogeneous catalysts on the yield of phenols, methoxyphenols and catechols from hydrothermal lignin degradation (Exp. 8-20 and 27-31), spruce-lignin/water = 133 g/L; T = 633 K; τ = 30 min; 0.10 g/glignin Raney-Ni, 0.08 g/glignin Rh on act. C, 0.08 g/glignin Pt on act. C, 0.07 g/glignin Co, 0.08 g/glignin PdCl2, 0.13 g/glignin Al-Ni; reactor pressurised with a) N2 (5 MPa initial pressure) and b) H2 (5 MPa initial pressure)...... 38 5.2 Yield of PhOH, methoxyPhOH and catechols at different loadings of Raney-Ni (Exp 10-11 and 21-31); spruce-lignin/water = 133 g/L; T =633 K; τ =30 min...... 40 5.3 Product composition after hydrogenation of 10 g/L catechol in water at different temperatures with 55 mg(dry matter) Raney-Ni (Exp 36-39); 1 MPa initial pressure H2; τ = 60 min. 40 5.4 Selectivity of phenol over catechol conversion after hydrothermal catechol decomposition catalysed with Raney-Ni (Exp. 103- 105, 114-118 and 122-125); input: 5 ml catechol/water-solution (5 g/L); different temperatures and residence times...... 41

6.1 Liquid, gas and solid reaction products at varying residence times are compared with the input amounts of Lig, EtOH and FA at reaction time zero of the experimental series. All experiments (Exp. 44-53) were conducted at the same loading conditions: 0.33 g Lig, 0.27 g FA, and 2.5 g EtOH; T = 633 K. 47 6.2 Individual yield of the gaseous product components, quanti- ﬁed onGC-FID andGC-TCD. The values for CO and H 2 have been multiplied by a factor of 10, respectively 10000, to enable a better visibility on the given scale; T = 633 K.... 47

134 LIST OF FIGURES

6.3 Gas and solid residue yields for experiments carried out at varying residence times. A transition zone, indicating the change from largely lignin characteristic solids to char is indicated, utilising the results from FT-IR analysis; T = 633 K.. 48 6.4 FT-IR spectra of unreacted lignin and of the collected solid residues at 90, 120 and 1180 min residence time. The largest alterations take place during the first 120 min and are documented in Table 6.1; T = 633 K...... 49 6.5 Yields of key components for the different phenolic compound classes: a) methoxyphenols, b) catechols and c) phenols over residence time; T = 633 K...... 52 6.6 Methoxyphenols, phenols, 4-ethylphenol and catechols in dependence from the conversion, a) first rank Delplot, b) second rank Delplot; T = 633 K...... 53 6.7 Scheme of main reaction participants and pathways of the depolymerisation of wheat straw lignin in ethanol and formic acid; gaseous products of ethanol and formic acid gasification take part in reactions 2-4, 7-9, 14 and 17. This is indicated by curved arrows; T = 633 K...... 56 6.8 The experimental values are compared with the model fit functions for the quantified key phenolic compounds; T = 633 K. 58 6.9 Fits of the gas-lump and the solid residue compared with the experimental data. The differentiation between lignin and char in the solid residue was accomplished largely by FT-IR analysis, as illustrated in Figure 6.4; T = 633 K...... 58 6.10 Comparison of calculated and experimental data for product yields of all phenolic (lump) components. Normalising factors (mg/glignin): MethoxyPhOH: 14.0, Catechols: 7.0, PhOH: 3.0, 4-EtPhe: 5.0, 2-EtPhe: 1.5...... 60 6.11 Results of lignin degradation inMA1 and CSTR at 653 K and 60 min (mean) residence times (Exp. 58, 63); a) liquid, gas and solid reaction products compared on a wt% basis with the input amounts of lignin from wheat straw Lig, EtOH and FA, b) Composition of the product gas, c) yields of PhOH, methoxyPhOH and catechols...... 64 6.12 Comparison between the CSTR-fit of phenolic lumps and their measured quantities for Exp. 64-66 conducted at a) 633 K, b) 653 K and c) 673 K; mean residence times are between 39 and 56 min (see Table 6.4)...... 66

135 LIST OF FIGURES

6.13 Comparison of the fluxes of each reaction defined in the model (see Figure 6.7) calculated for both batch-fit and CSTR-fit at 60 min (mean) residence time τ (τ 0) and 633 K (see Ap- pendix C.6)...... 69 6.14 Sensitivity on a scale from 0 to 100 % (see Appendix C.5) of the calculated yields at 633 K and 60 min (mean) residence time to a variation of rate coefficients by factor 2 for a) batch- fit and b) CSTR-fit...... 71 6.15 Results from the CSTR-fit of the experimental yields at different temperatures and mean residence times (Exp. 64-66) of a) methoxyphenols and b) catechols...... 72 6.16 Comparison of the normalised calculated and experimental data for product yields of all phenolic (lump) components. The following normalising factors (mg/glignin) were used for better comparability: MethoxyPhOH: 42.1; Catechols: 20.4; PhOH: 5.6; 4-EtPhe: 19.4; 2-EtPhe: 3.1...... 74 6.17 Scheme of main reaction participants and pathways of the hydrothermal depolymerisation of spruce-lignin; gaseous components take part in reactions 2-6 and 13. This is indicated by curved arrows...... 77 6.18 Comparison between the model-fit of solid and gaseous lump components and measured quantities of solid residue and stable gaseous compounds (CO2 and CH4) for experiments carried out at a) 593 K, b) 613 K, c) 633 K, d) 653 K...... 80 6.19 Comparison between the model-fit of phenolic lumps and their measured quantities for experiments conducted at a) 593 K, b) 613 K, c) 633 K, d) 653 K; error bars show standard deviation σ obtained from similar reproduced experiments at the corresponding residence time and temperature...... 81 6.20 Fluxes of main reactions comprised by the developed model of hydrothermal depolymerisation of spruce-lignin (see Fig- ure 6.17), reactions r1, r2, r4, r5, r6, r7, r9, r11 are compared with analogue reactions of lignin solvolysis in ethanol (see Section 6.1); τ = 131 min; T = 633 K; lignin/water=132 g/L. 84 6.21 Sensitivity matrices considering comparable reactions and intermediates or products of a) the model of hydrothermal degradation of spruce-lignin and b) the solvolysis in ethanol and formic acid of wheat-straw-lignin; τ = 131 min; T = 633 K; lignin/solvent=132 g/L...... 85

136 LIST OF FIGURES

7.1 Process scheme with main components (labelled with letters) an streams (labelled with numbers)...... 95 7.2 COM of phenol from lignin for the basic case and 2 diﬀerent scenarios (see Table 7.10) for a) the water approach and b) the ethanol approach...... 108 7.3 COM of phenol from lignin in dependence of the phenol yield for the basic case of both ethanol and water approach; scenario 3 (see Table 7.10) is displayed for the water approach..... 108

A.1 Mechanical drawing of theMA1, simple version, 10 mL inner volume...... 142 A.2 Mechanical drawing of theMA2, gas-input version for gas ventilation and loading, 5 mL inner volume...... 143 A.3 Schematic of the CSTR including feeding, reactor and collection.144

B.1 Sample workup scheme for product phases of experiments with ethanol...... 146 B.2 sample workup scheme for product phases of experiments with water...... 146

C.1 Mass balanced reactions for modelling thermal degradation of wheat-straw lignin in ethanol...... 155 C.2 Mass balanced reactions for modelling the hydrothermal degradation of spruce-lignin...... 158

D.1 Comparison of homogeneous basic catalysts with 0.12 mol/L input concentration in a spruce-lignin/water = 133 g/L suspension; T = 633 K (Exp. 1 and 7-9); τ = 30 min...... 164 D.2 Comparison of homogeneous acidic catalysts with varying input concentration in a spruce-lignin/water = 133 g/L suspension (Exp. 2-6 and 8-9); T = 633 K; τ = 30 min...... 164 D.3 Mass fractions of the input (residence time = 0) and output at 573 K and diﬀerent residence times after hydrothermal decomposition of catechol in a 10 mL-MA2; 5 ml input of aqueous solution (catechol/water = 10 g/L); 2.2 mg Raney-Ni catalyst 166 D.4 Reaction scheme of the catalytic decomposition of catechol in water at temperatures between 523 K and 573 K based on delplot-analysis and literature research...... 167

137 LIST OF FIGURES

D.5 Concentration of catechol in the product after hydrothermal decomposition of catechol depending on the residence time at a) 573 K, b) 533 K, c) 523 K, d) 513 K; Data points represent experimental values, continuous line calculated according to Equation D.1; 5 ml input of catechol/water = 10 g/L; 2.2 mg Raney-Ni catalyst...... 168 D.6 Concentration of phenol in the product after hydrothermal decomposition of catechol depending on the residence time at a) 513 K, b) 523 K, c) 533 K and d) 573 K; Data points from experiments, continuous line calculated according to Equa- tion D.2; 5 ml input of catechol/water = 10 g/L; 2.2 mg Raney-Ni catalyst...... 169 D.7 Arrhenius plot of pseudo-ﬁrst order rate coeﬃcients of the hydrothermal Raney-Ni catalysed decomposition of catechol (circular data points) and phenol (rectangular data points), k0 = 1min−1 ...... 170

138 Appendix A

Materials

In this appendix the chemicals used for synthesis or analysis and their degree of purification are listed (see Table A.1). All chemicals were used as purchased. No further purification was required. The catalysts are listed sep- arately in Table A.2 including the absorption surface according to Brunauer, Emmett, Teller (BET)-analysis. Furthermore, the composition of the alloys for the reactors are listed, a mechanical drawing of theMA and a flow scheme of the CSTR are provided.

139 Table A.1: Chemicals used

Chemical Grade Producer Ethanol ≥ 99.5% Merck FA 98-100% Merck acetic acid (AA) 96% Merck Ethyl acetate ≥ 99.5% Fluka Water ≥ 18.4 MΩ·cm Millipore Direct Q3 KOH acidimetric assay ≥ 85% Merck NaOH ≥ 99% Anala R HCl 37% Merck

H2 ≥ 99.999% Air Liquid N2 ≥ 99.9999% Air Liquid Pentacosane ≥ 99% Alfa Aesar Pentadecane ≥ 99% Aldrich KBr ≥ 99% Merck Phenol ≥ 99% Merck o-Cresol ≥ 99% Merck p-Cresol ≥ 98% Merck 2-Ethylphenol ≥ 98% Merck 4-Ethylphenol ≥ 98% Merck Guaiacol ≥ 98% Merck 4-Methylguaiacol ≥ 98% Merck 4-Ethylguaiacol ≥ 98% Alfa Aesar Syringol ≥ 99% Aldrich Catechol ≥ 99% Alfa Aesar 4-Methylcatechol ≥ 96% Alfa Aesar 4-Ethylcatechol ≥ 98% Alfa Aesar Cyclohexanol ≥ 99% Merck Cyclohexanone ≥ 99% Merck

PdCl2 ≥ 99% Merck

140 Chapter A. Materials

Table A.2: Catalysts

Catalyst Carrier- Grade BET-surface Producer Material Raney-Ni None 50 wt% in 53 m2/g Merck

H2O Pd ActivatedC 10 wt% 900 m 2/g Fluka 2 Pt Al2O3 5 wt% 12 m /g Merck Rh ActivatedC 5 wt% 900 m 2/g Merck Al-Ni 50 wt% Ni, 4 m2/g Aldrich 50 wt% Al Co None ≥ 99.6 wt% 23 m2/g Merck

Table A.3: Construction materials

Material Composition max. (min) wt% Inconel625 Ni (58) chromium (Cr) 20-30 Mo 8-10 niobium (Nb)& tantalum (Ta)3.15-4.15 iron (Fe)5 1.4571C 0.03 silicon (Si) 1.0 Mn 1.5 phosphor (P) 0.04S 0.015 Cr10.5-12.5 Mo 2.0-2.5 Ni 10.0-13.5 titanium (Ti) 5xC- 0.7

141 142

1 2 3 4 5 6 7 8

A A

B B

C C

D D

E E

EDV Nr. FSCM-Nr. Blatt 0DVWDE Miniautoklav10ml 1 1:1 Format A3

Bearb. 29.09.2009 Pagel

Gepr.

F Best. Miniautoklav F

Ausg. 10ml Zeichn.-Nr. Zust. 1.4571 Vertr.-Nr. - 1 2 3 6 7 8

Figure A.1: Mechanical drawing of theMA1, simple version, 10 mL inner volume. Chapter A. Materials Figure A.2: Mechanical drawing of the MA 2, gas-input version for gas ventilation and loading, 5 mL inner volume.

143 144

PI EtOH Feed Screw press 1

Screw PI press 2 TIC

el. +

Gaseous - products Reactor Flash PIC Separator

Liquid & solid Cooling products water

Figure A.3: Schematic of the CSTR including feeding, reactor and collection. Appendix B

Sample workup

This appendix provides more detailed information about the workup of the solid, liquid and gaseous samples obtained from the experiments. The Sec- tion B.1 provides an overview for the separation of the different product phases and the analytical methods applied on them. In Section B.2 the extraction of the phenolic products from the aqueous liquid phase is explained in detail including the distribution coefficients of the phenolics which were quantified byGC-FID.

B.1 Analysis overview

After the separation of the gaseous phase which is different for either CSTR orMA, the gas was analysed on aGC equipped with both, a FID and a TCD. The mixture of liquid and solid residues was filtered with a filter purchased from Pall with 45 nm pore size for the separation of solid and liquid phase. The solid fraction was weighed, dried for 16 h at 378 K and weighed again after drying. On the dried solid FT-IR and elementary analysis is applied. In the case water was used as solvent, the liquid products had to be extracted from the aqueous sample with ethyl acetate. After extraction the organic phase can be analysed withGC-FID or eventuallyGC-MS exactly like the liquid product from experiments with ethanol. For that purpose the organic sample, containing the liquid products diluted in ethyl acetate or ethanol, was mixed with an equal amount of ethyl acetate containing the internal standard pentacosane. Figure B.1 and B.2 show the schematic sample workup for experiments with ethanol and water, respectively.

145 B.1. Analysis overview

Products

Flash Separation

Gas Liquid and Solid

Volumetric Filtering GC-FID/TCD Analysis Liquid+EtOH- Solid+EtOH- soluble insoluble

GC-FID + GC-MS Drying

Karl-Fischer- Gravimetric IR Phenolics Titration Analysis

Figure B.1: Sample workup scheme for product phases of experiments with ethanol.

Products

Flash Separation

Gas Liquid and Solid Filtering

GC-FID/TCD Volumetric Liquid + Solid + H2O- Analysis H2O- soluble insoluble

Extraction Drying Organic phase Aqueous phase

Gravimetric GC-FID + GC-MS FT-IR UV-VIS HPLC Analysis

Figure B.2: sample workup scheme for product phases of experiments with water.

146 Chapter B. Sample workup

B.2 Extraction

For the analysis of phenolic products from the hydrothermal lignin degradation viaGC, phenolics had to be extracted from the aqueous product phase. Ethyl acetate was selected as extractant. For a better performance of the extraction the pH of the aqueous phase was adjusted to a value between 3 and 5 by adding an adequate amount of HCl. 1.3 mL of the sample were thereafter mixed with 0.52 mL ethyl acetate. The phases were left to settle for one hour. A sample of the organic phase was analysed on aGC equipped with a FID, optionally onGC equipped with aMS. The concentration of the diﬀerent compounds (solutes) in the original aqueous product phase can be calculated with the following Equation B.1

fa · fb Ci,aq,0 = Ci,org · , fa = 2, (B.1) fc · fd,i

where Ci,org is the measured concentration of the (solute) in the organic phase, Ci,aq,0 the concentration in the aqueous product phase before extraction and fa, fb, fc and fd,i are factors considering the performance of the extraction. The factor fa considers the dilution of the organic phase after the extraction with ethyl acetate including the internal standard pentacosane (C25H52) for the preparation ofGC-analysis (ratio sample:standard = 1:1). The solvent (water) and extractant (ethyl acetate) are not completely insoluble. This fact was considered by the factor fb which is proportional to the miscibility of ethyl acetate in water. The latter was found to be independent from the solute and its concentration. It was determined by adding the extraction standard pentadecane (C15H32) to the extractant which is practically insoluble in water. The concentration of pentadecane in the organic extractant was measured before (CC15H32,org,0) and after (CC15H32,org) the extraction. With the quotient of these values fb can be calculated according to Equation B.2. The average value for fb was 0.815 with a standard deviation of 0.024.

CC15H32,org,0 fb = = 0.815 ± 0.024 (B.2) CC15H32,org

The factor fc considers the alteration of the solute concentration due to the diﬀerent input volumes of aqueous sample (Vaq) and extractant (Vorg) and is calculated according to the following Equation B.3

Vaq 1.3 mL fc = = = 2.5. (B.3) Vorg 0.52 mL

147 B.2. Extraction

The factor fd,i considers the distribution ratio of the solutes i in both, the solvent and the extractant. It has to be clearly distinguished from the distribution coeﬃcient. In fact, fd,i seems most convenient for the calculation of Ci,aq,0 as it relates the concentration of the solute in the extractant after the extraction mi,org with the initial concentration in the aqueous phase mi,aq,0.

mi,org fd,i = (B.4) mi,aq,0

The experimental evaluation of this factor fd,i and a list of the factors for each solute i are demonstrated in the following. For the evaluation of the factor fd,i several extraction experiments were carried out. A single solute (single experiments) or a mixture of various solutes (bulk experiments) were diluted in the solvent water with a determined concentration Ci,aq,0. Eventually NaCl was added in order to evaluate the inﬂuence of salts on the distribution between the two phases. The employed extractant was ethyl acetate containing 1000 mg/L pentadecane. The extraction was carried out as described above. The results are listed in Table B.2. Extraction experiments with phenol, guaiacol, catechol, cyclopentanone, 2-methylcyclopentanone and cyclohexanol show good reproducibility even in the presence of other phenolics in the solvent. However, high amounts of salt (NaCl) and low solute concentrations give low values of fd,i and an increased standard deviation. Considering the ash content of the diﬀer- ent types of lignin, a maximum salt concentration between 700 mg/L and 3000 mg/L is expected. However, increased salt concentrations in combination with low solute concentrations lead to lower values of fd,i, that is an underestimation of the compound concentration in the aqueous phase and the compound yield. Thus, an assumption of a high value for fd,i must be legitimate. Extraction of all ethylated phenolics, namely 2-ethylphenol, 4- ethylphenol, 4-ethylguaiacol and 4-ethylcatechol, yield a fd,i between 80 % and 90 %. Although the standard deviation is relatively high (above 5 %), the assumption of 80 % < fd,i < 90 % seams reasonable. The increased uncertainty arising from the extraction of course has to be considered for all yields given within this work. High standard deviations and bad agreement for single and bulk experiments are obtained for cyclohexanone. Thus, fd,i for cyclohexanone is set to the highest measured value (80 %) with a standard deviation of 20 %. The distribution of methylcatechol, methylguaiacol, p-cresol and o-cresol can be determined with low standard deviation by UV-Vis. UV-Vis can however only be applied on the solvent after single experiments, that is, the obtained value does neither take into account the matrix in the aqueous phase, nor the analysis method which is usually

148 Chapter B. Sample workup applied on the samples. In the case of methylcatechol, methylguaiacol and o-cresol the determination of fd,i viaGC-analysis leads to varying values and high standard deviations. The value obtained from UV-Vis-analysis is employed for further calculation combined with the maximum standard deviation resulting from allGC-analysis-results. On the basis of the conducted experiments and the observations made above, the following values for fd,i are applied for the calculation of the concentration in the aqueous phase (see Table B.1).

Table B.1: Applied values of fd,i for quantiﬁed phenolics i.

Compound Ci,aq,0 Standard Applied deviation fd,i mg/L % % 2-Ethylphenol 500-1000 6.6 89.0 2-Methylcyclopentanone 200-2000 3.6 77.4 4-Ethylcatechol 500-1000 4.6 88.7 4-Ethylguaiacol 500-1000 5.2 82.2 4-Ethylphenol 500-1000 6.7 85.0 4-Methylcatechol 500-1000 5.9 88.6 4-Methylguaiacol 250-1000 8.1 87.5 Catechol 200-2000 4.1 83.2 Cyclohexanol 100-1000 9.3 67.3 Cyclohexanone 200-2000 20.0 80.0 Cyclopentanone 200-2000 3.6 53.9 Guaiacol 100-1000 1.7 93.5 o-Cresol 500-2000 14.0 96.5 p-Cresol 200-1000 6.2 98.1 Phenol 200-1000 1.6 94.5

149 B.2. Extraction

Table B.2: Distribution of phenolic products in a two phase system comprising water and ethyl acetate under diﬀerent conditions.

Compound Matrix Analysis Ci,aq,0 NaCl Exp. Standard Average method redun- devia- fd,i dancy tion mg/L mg/L % % 2-Ethyl- bulkGC 500-1000 0 8 6.6 89.0 phenol bulkGC 100-1000 1000 8 5.3 86.0 2-Methylcyclo- singleGC 500-2000 0 16 3.0 77.4 pentanone singleGC 1000 10000 7 1.3 77.7 bulkGC 200-2000 0 14 3.6 76.5 bulkGC 100-2000 1000 8 6.9 71.0 4-Ethyl- bulkGC 500-1000 0 8 4.6 88.7 catechol bulkGC 100-1000 1000 8 9.5 75.6 4-Ethyl- bulkGC 500-1000 0 8 5.2 82.2 guaiacol bulkGC 100-1000 1000 8 4.1 79.8 4-Ethyl- bulkGC 500-1000 0 8 6.7 85.0 phenol bulkGC 100-1000 1000 8 5.0 82.2 4-Methyl- singleGC 500-2000 0 18 5.5 83.7 catechol singleGC 1000 10000 6 2.2 95.6 single UV-Vis 500-1000 0 12 1.2 88.6 bulkGC 100-1000 0 8 3.0 78.2 bulkGC 500-1000 1000 6 2.9 71.7 4-Methyl- singleGC 500-2000 0 18 7.1 111.3 guaiacol singleGC 1000 10000 6 2.7 99.8 single UV-Vis 250-1000 0 12 1.1 87.5 Catechol singleGC 500-1500 0 10 2.0 83.5 bulkGC 200-2000 0 21 4.1 83.2 bulkGC 100-1000 1000 8 5.8 75.9 Cyclo- singleGC 500-2000 0 18 2.7 69.0 hexanol singleGC 1000 10000 6 2.5 63.8 bulkGC 100-1000 0 8 1.6 67.3 bulkGC 50-1000 1000 9 9.3 58.4 Cyclo- singleGC 500-2000 0 18 4.4 57.5 hexanone singleGC 1000 10000 6 2.3 61.1 bulkGC 200-2000 0 11 3.2 80.0 bulkGC 100-2000 1000 8 6.8 73.1 Cyclo- singleGC 500-2000 0 18 2.3 52.8 pentanone singleGC 1000 10000 6 1.2 56.9 bulkGC 200-2000 0 19 3.6 53.9 bulkGC 100-2000 1000 8 5.9 46.9 Guaiacol bulkGC 100-1000 0 8 1.7 93.5 bulkGC 50-1000 1000 9 12.1 82.3 singleGC 500-2000 0 8 1.7 90.9 singleGC 1000 1000 4 1.2 91.3 o-Cresol bulkGC 200-1000 0 6 15.9 102.3 bulkGC 100-1000 1000 11 10.1 88.7 singleGC 500-2000 0 18 13.3 102.4 singleGC 1000 10000 6 4.1 111.9 single UV-Vis 500-2000 0 12 0.9 96.5 p-Cresol bulkGC 200-1000 0 11 6.2 98.1 bulkGC 100-1000 1000 8 5.5 93.6 singleGC 500-2000 0 18 3.4 85.5 single UV-Vis 500-2000 0 12 0.4 96.6 Phenol bulkGC 100-1000 0 8 1.6 94.5 bulkGC 50-1000 1000 9 5.8 88.1 singleGC 1000 0 6 1.9 98.2 singleGC 1000 1000 2 3.8 96.8

150 Appendix C

Modelling

This appendix contains detailed information about the equations and methods used for the modelling of the formal reaction kinetics of thermal lignin degradation in both solvents, ethanol and water. Information about the central optimisation function can be found in Section C.1. All mass balanced reactions and the resulting differential equations which were used for modelling, the thermal degradation of lignin originating from wheat straw in ethanol and formic acid are shown and explained in Section C.2. All mass balanced reactions and the resulting differential equations which were used for modelling the thermal degradation of lignin originating from spruce in water are shown and explained in Section C.3. In Section C.5 the formula for the calculation of the sensitivity analysis is displayed and explained. In Section C.6 the formula for the calculation of the flux analysis is displayed and explained.

C.1 Optimisation function

Modelling was performed in Matlab V 7.12.0, MathWorks using a set of differential equations, which were formulated according to the developed model of reaction participants and pathways which depends mainly on the employed type of solvent and the type of lignin. Every model comprises a number of l defined (lump) components and n reaction pathways. For each reaction pathway a formal kinetic rate coefficient is defined. The reaction order for all reactions is set to one. The unconstrained non-linear Matlab- function «fminsearch«, which uses the simplex search method by Lagarias et al. [Lag+98], was used for optimisation, because of its robustness and

151 C.1. Optimisation function because it does not need derivatives of the objection function. The function "fminsearch" ﬁnds the minimum of the scalar function given in Equation C.1, starting at kj = 0.001:

l ~ X X meas calc ~ f(k) = exp(wi[yi,ξ − yi (k)])+ i=1 ξ (C.1) " l l !# X calc ~ X exp wl+1 yi (k) − yi,0 , i i

where yi is the yield deﬁned by the following Equation C.2

mi yi = , (C.2) mLig,0 i indicates a reaction participant, ξ indicates an experiment, 0 indicates an input parameter, Lig indicates the reactant lignin, m is the mass, w is the weighing factor and ~k ∈ Rn comprises the formal kinetic rate coeﬃcients of all reactions kj. The function f (see Equation C.1) consists of a term which considers the diﬀerence between calculated and experimental yields and an additional term which forces the calculation to respect the overall mass balance. If the temperature (T ) dependence of kj of each reaction j was considered, the following Arrhenius-Equation C.3 was employed

E − A,j −1 kj(T ) = Aj · e RT ,R = 8.314 J/(mol · K ). (C.3)

The target values for optimisation in this case were EA,j and Aj. The initial values were chosen to be EA,j=1,2,...,n = 100 kJ/mol and Aj=1,2,...,n = 7 −1 10 min . If the temperature dependence of kj was not considered, as for example in Section 6.1, the target value of optimisation was ~k. The initial −3 −1 value was kj=1,2,...,n = 10 min . The term «apparent activation energy EA,j» which expresses the temperature dependence of the formal kinetic rate coefficient kj is used throughout this work. Since EA,j should fulfil Equation C.3, which comprises the ideal gas constant R, it necessarily has the unit J/mol. This is not correct from a strict scientific point of view, since most reactants are not ideal gases. How- ever, the term «activation energy» is predominantly used in literature, even if only solid lignin is involved in the reaction. Thus for better comparison of the data obtained here with values obtained by other authors, the temperature dependence of the reactions was expressed by the term EA,j. It should further be pointed out that a chemical reaction model would only then be

152 Chapter C. Modelling able to describe the experimental data correctly, if all the relevant reaction pathways are contained in the model. The ultimate number of reactions (and rate coeﬃcients kj) in a model are of minor signiﬁcance.

C.2 Solvolysis in ethanol

The Equations (C.4) to (C.14) are the diﬀerential equations describing the formal kinetics of the developed model for thermal lignin degradation in ethanol and formic acid. They are based on the mass-balanced reactions shown in Figure C.1. The index j of the formal kinetic parameters kj corresponds to the index of the reactions displayed in Figure C.1. Each index i of the yields yi corresponds to one participating component. All components and their indices are listed in Table C.1.

The differential equations describe the consumption and formation of all components i through the reactions j considered by the model for solvolysis of wheat straw lignin in ethanol and formic acid (see Figure 6.7), where yi,0 and yi are, respectively, the entering and outgoing mass of the component i divided by the input mass of lignin mLig,0. If the model is applied for the simulation of thermal lignin degradation in a batch reactor Equations (C.4) m˙ to (C.14) were used without the term (yi,0 − yi) · m . If it is used for the simulation of thermal lignin degradation in a CSTR they have to be employed with the mentioned term in rectangular brackets. The latter considers the continuous character of the process and the mixed flow inside the CSTR, assuming an ideal reactor. The following assumptions are to be considered only for CSTR: The quotient of the flow rate m˙ divided by the reaction mass m is equivalent to the mean residence time τ 0. The reaction mass is estimated by the volume of the reactor (200 mL) and the density of the solvent at given temperature and pressure displayed in Table 6.4. The experimental duration is defined as t, the set-point t = 0 is the start of the experiment when the suspension of ethanol, formic acid and lignin is pumped into the reactor.

153 C.2. Solvolysis in ethanol

dy m˙ 1 = − k y − k y − k y + (y − y ) · (C.4) dt 1 1 6 1 10 1 1,0 1 m dy m˙ 2 =k y + k y − k y + (y − y ) · (C.5) dt 1 1 15 8 2 2 2,0 2 m dy m˙ 3 =0.89 · k y − k y − k y + (y − y ) · (C.6) dt 2 2 4 3 3 3 3,0 3 m dy 4 =0.85 · k y + 0.77 · (k y − k y + k y − k y )+ dt 4 3 7 6 14 4 9 7 8 4 (C.7) m˙ − k y − k y + (y − y ) · 5 4 17 4 4,0 4 m dy m˙ 5 =0.91 · k y + 0.98 · k y + k y + (y − y ) · (C.8) dt 3 3 17 4 16 8 5,0 5 m dy m˙ 6 =k y − k y + k y + (y − y ) · (C.9) dt 6 1 7 6 14 4 6,0 6 m dy m˙ 7 =k y − k y + (y − y ) · (C.10) dt 8 4 9 7 7,0 7 m dy m˙ 8 =k y − k y − k y − k y + (y − y ) · (C.11) dt 10 1 11 8 15 8 16 8 8,0 8 m dy 9 =0.11 · k y + 0.09 · k y + 0.15 · k y + k y + dt 2 2 3 3 4 3 5 4 0.23 · (k7y6 − k14y4 + k9y7 − k8y4 + k11y8) + k12y11+ (C.12) m˙ k y + 0.02 · k y + (y − y ) · 13 10 17 4 9,0 9 m dy m˙ 10 = − k y + (y − y ) · (C.13) dt 13 10 10,0 10 m dy m˙ 11 = − k y + (y − y ) · (C.14) dt 12 11 11,0 11 m

For the kinetic modelling pseudo-elementary reactions were defined (see Figure C.1). In fact, the reactions r2 - r4 and r17 involve lumped components which can not be associated with a molecular structure as indicated in the reaction equation. However, the defined mass balanced reactions provide an orientation for the mass distribution throughout the reaction. The employed coefficients might be reconsidered in the future respecting the important condition that the right and the left side of each equation are equilibrated.

154 Chapter C. Modelling

OH OH O +H OH CH3 2 0.89 + 0.11 Gas k 2

OH OH OH +H 2 0.85 + 0.15 Gas k 4

OH OH OH OH OH O O 0.91 + 0.09 Gas k 3 n

OH OH O O 0.98 + 0.02 Gas k 17 n

OH OH k CH3 9 0.77 + 0.23 Gas k 8

OH OH k 7 0.77 + 0.23 Gas k 14

H3C

Figure C.1: Mass balanced reactions for modelling thermal degradation of wheat-straw lignin in ethanol.

Table C.1: Indices of reaction participant Index Participant Index Participant 1 Lignin 7 2-Ethylphenol 2 methoxyPhOH 8 LD, EtOH-soluble residue 3 Catechols 9 Gas 4 PhOH 10 Formic Acid 5 Char, EtOH-insoluble residuea 11 Ethanol 6 4-Ethylphenol a except not reacted lignin.

155 C.3. Solvolysis in water

C.3 Solvolysis in water

The Equations (C.15) to (C.24) are the diﬀerential equations describing the formal kinetics of the developed model for thermal lignin degradation in water. They are based on the mass-balanced reactions shown in Figure C.2. The index of the formal kinetic parameters corresponds to the index of the reactions displayed in Figure C.2. Each index i of the yields yi or masses mi corresponds to one participating component i. All components and indices are listed in Table C.1.

The differential equations describe the consumption and formation of all components i through the reactions j considered by the model for hydrothermal degradation of spruce lignin (see Figure 6.17), where yi,0 and yi are, respectively, before and after the reaction the mass of the component i divided by the input mass of lignin mLig,0. The optimisation function uses a numerical solution of Equations (C.15) to (C.24) for the calculation of yi. For the kinetic modelling pseudo-elementary reactions were defined (see Figure C.1). In fact, the reactions r2 - r6 involve lumped components which can not be associated with a molecular structure as indicated by the reaction equation. However, the defined mass balanced reactions provide an orientation for the mass distribution throughout the reaction. The employed coefficients might be reconsidered in the future respecting the important condition that the right and the left side of each equation are equilibrated.

156 Chapter C. Modelling

Table C.2: Indices of reaction participant Index Participant Index Participant 1 Lignin 6 LD2, reactive intermediates, not detectable inGC

2 methoxyPhOH 7 CO2 and CH4 3 Catechols 8 Water 4 PhOH 9 LD2, stable products, partially water-soluble a 5 Char, Water-insoluble residue 10 CO and H2 a except not reacted lignin.

dy 1 = − k y − k y (C.15) dt 1 1 7 1 dy 2 = k y − 0.98 · k y − k y (C.16) dt 1 1 2 2 3 2 dy 3 = 0.87 · k y + k y − 0.8 · k y − k y − k y (C.17) dt 2 2 10 6 4 3 5 3 12 3 dy 4 = 0.68 · k y + 0.76 · k y − 0.62 · k y (C.18) dt 4 3 3 2 6 4 dy 5 = 0.82 · k y + k y (C.19) dt 5 3 9 6 dy 6 = k y − (k + k + k + k + k )y (C.20) dt 7 1 8 9 10 11 15 6 dy 7 = 0.13 · k y + 0.24 · k y + 0.32 · k y + k y + dt 2 2 3 2 4 3 11 6 (C.21) + 0.96 · k13y10 + k14y10 + k6y4 dy 8 = 0.16 · k y − 0.35 · k y − 0.39 · k y (C.22) dt 5 3 6 4 13 10 dy 9 = k y + k y (C.23) dt 8 6 12 3 dy 10 = 0.02 · k y + k y − 0.02 · k y − 0.2 · k y − dt 5 3 15 6 2 2 4 3 (C.24) + 0.03 · k6y4 − 0.57 · k13y10 − k14y10

157 C.3. Solvolysis in water

OH OH O OH CH3 0.98 + 0.02 H 0.87 + 0.13 CH 2 k 4 2

OH OH O CH3 0.76 + 0.18 CO + 0.06 CH k 2 4 3

OH OH OH 0.8 + 0.2 CO 0.68 + 0.32 CO k 2 4

OH OH OH OH OH O O 0.82 + 0.16 H O + 0.02 H k 2 2 5 n

0.62 + 0.35 H O + 0.03 H 0.58 CO + 0.42 CH 2 2 k 2 4 6

0.39 H O + 0.61 CO 0.96 CO + 0.04 H 2 k 2 2 13

0.07 H + 0.93 CO 0.73 CO + 0.27 CH 2 k 2 4 14

Figure C.2: Mass balanced reactions for modelling the hydrothermal degradation of spruce-lignin.

158 Chapter C. Modelling

C.4 Analysis of error

For the analysis of the experimental error, multiple identical experiments were carried out. For the evaluation of the experimental error the standard deviation σ was employed. For the evaulation of the quality of kinetic modelling the coefficient of determination R2 and the standard deviation σ were employed. R2 was calculated (see Equation C.25) by comparing the differ- calc meas ence of calculated yield yi or concentration and the measured yield yi,ξ meas or concentration with the difference of measured yield yi,ξ or concentration meas and the average yield yi or concentration.

P (ycalc − ymeas)2 P (Ccalc − Cmeas)2 R2 = 1 − ξ i i,ξ = 1 − ξ i i,ξ , (C.25) P (ymeas − ymeas)2 P meas meas 2 ξ i i,ξ ξ(Ci − Ci,ξ )

meas where yi is the average of the yield of the component i measured in the product of all experiments carried out at the same temperature. The standard deviation σ is given by Equation C.26 v u Ξ calc meas !2 u X yi − yi,ξ σ = t(Ξ − 1)−1 , (C.26) ymeas ξ=1 i,ξ where Ξ is the number of experiments which are considered. The dif- calc meas ference between calculated yield yi and measured yield yi,ξ is normed meas by dividing by yi,ξ (see Equation C.26). If the standard deviation of the results of several identical experiments was calculated, glssymb:yicalc is to be replaced by the average of all considered identical experimental results yi. The denominator of the squared term is to be replaced by the average value yi (see Equation C.27). The resulting value standard deviation is mostly displayed in %, thus multiplied by the factor 100. v u Ξ meas 2 u X yi − yi,ξ σ = t(Ξ − 1)−1 , (C.27) y ξ=1 i

C.5 Sensitivity analysis

The sensitivity analysis shows the influence of the different reactions defined in a reaction model on the yield of the components. Single rate coefficients are altered by a factor of 2 and the resulting yields of the components are

159 C.6. Flux analysis compared with the original values. This difference is defined as the sensitivity. The sensitivity of a each component i to each reaction rj is displayed in the sensitivity matrix (see Figure 6.14 and 6.21). In the case of the batch experiments the sensitivity ∆yi,Batch is calculated by integration of the dif- 0 ference between the original yield yi and the yield yi after modification of kj. The lower limit of integration is the start of the reaction t0 and the upper limit is the determined residence time τ (see Equation C.28). Contemplat- ing Equation C.28 it becomes clear that the choice of the upper integration limit changes the resulting sensitivity. For comparison of the batch-fit and the CSTR-fit, the upper integration limit for ∆yi,Batch is chosen to be equal to the mean residence time in the CSTR τ = τ 0.

Z τ 0 1 yi − yi ∆yi,Batch = · dt, (C.28) τ t0 yi

In case of the CSTR the sensitivity is quantified by ∆yi,CSTR , which is the difference between the original yield yi and the yield after modification of 0 each rate coefficient yi in the stationary state (t → ∞) at a set temperature and mean residence time τ 0.

0 yi − yi ∆yi,CST R = (C.29) yi

C.6 Flux analysis

The ﬂux implies the mass of a component i that disappears via a certain reaction rj during a certain time period in a determined reaction volume. In the case of the batch reactor, the ﬂux from the start of the reaction until the residence time τ was calculated by integration, according to Equation C.30:

Z τ fluxj,Batch = CLig,0 · kjyi,j dt, (C.30) t0

where CLig,0 is the initial concentration of lignin in the batch autoclave, kj the formal kinetic rate coefficient of the first order reaction rj and yi,j the time depending yield of the corresponding component i which is either a reactant or a product of this reaction. Since reactions are assumed to be first order reactions, the flux of each reaction depends on the yield yi,j of exactly one corresponding component. For comparison of the the fluxes in the two different types of reactor the integration limit τ was chosen to be identical to the mean residence time in the CSTR τ 0. In the case of the CSTR the

160 Chapter C. Modelling

ﬂux in the steady state is calculated according to Equation C.31:

0 fluxj,CST R = CLig,0 · τ · kjyi,j, (C.31) 0 where τ is the mean residence time and yi,j is independent from the experimental duration since stationary state is assumed (t → ∞).

161 C.6. Flux analysis

162 Appendix D

Extended studies of catalysts

D.1 Homogeneous catalysts

Homogeneous catalysts in water were tested at a reaction temperature of 633 K and a residence time of 30 min. Basic catalysts, that are KOH and NaOH, as well as acidic catalysts, that are HCl,AA and FA, were used. Basic catalysts promote the formation of catechols. Their yield increases from approximately 5 to approximately 20 mg/glignin. The yield of methoxyphenols decreases slightly and the yield on phenols increases (see Figure D.1).

The use of acidic catalysts give very different results. If HCl is employed as catalyst for the hydrothermal degradation of spruce-lignin the yields of the main products are strongly dependent on the concentration of HCl. Lower concentrations (0.33 mol/L) yield relatively high amounts of catechols (16 mg/glignin) and CO (29 mg/glignin). Methoxyphenols could not be detected neither at low nor high concentrations (1.81 mol/L) of HCl. The yield of catechols and CO significantly decrease when the concentration of HCl is increased to 1.81 mol/L to approximately 10 mg/glignin and 12 mg/glignin respectively (see Figure D.2). In addition, at increasing concentrations of HCl an increase of the yield of CO2 and hydrocarbon (HC)- gases is observed. The addition ofAA into the reactor has no or a very small effect on the product distribution.

163 D.1. Homogeneous catalysts

N o n e N a O H 2 0

) K O H n i n g i l g / g

m 1 0 (

d l e i Y 0 P h O H M e t h o x y P h O H C a t e c h o l s

Figure D.1: Comparison of homogeneous basic catalysts with 0.12 mol/L input concentration in a spruce-lignin/water = 133 g/L suspension; T = 633 K (Exp. 1 and 7-9); τ = 30 min

9 0 8 0 0 a ) N o C a t a l y s t s b ) H C l ( 0 . 3 3 m o l / L )

) H C l ( 1 . 8 1 m o l / L ) n i

n 6 0 g

i A A ( 0 . 5 8 m o l / L ) l g

/ A A ( 2 . 9 2 m o l / L ) 4 0 0 g m (

3 0 d l e i Y 0 0 r H H l s e s O O 2 a h O h O h o a s C C C h P y P e c G x a t C - t h o C H M e

Figure D.2: Comparison of homogeneous acidic catalysts with varying input concentration in a spruce-lignin/water = 133 g/L suspension (Exp. 2-6 and 8- 9); T = 633 K; τ = 30 min

164 Chapter D. Extended studies of catalysts

HCl obviously has a strong effect not only on the cleavage of the bonds between lignin monomers, as the increased yield of catechols shows, but also on the gasification and the char yield. Low concentrations yield high amounts of CO, probably coming from the lignin structure due to increased cleavage of ether bonds connected with the liberation of CO. Increased amounts of HCl lead to very high yields of CO2 and Char. Water-gas shift reaction or oxidation of CO might be responsible for increased amounts of CO2. The effects of formic acid on the phenolic products observed at the mentioned temperature are weaker than those of HCl. Obviously the input of acetic acid has to be increased significantly in order to obtain a similar effect as contemplated for experiments with HCl. A catalytic effect of both, acidic and basic media, on the hydrolysis of lignin in near critical water was observed like suggested by Toor, Rosendahl, and Rudolf [TRR11]. A more detailed investigation in order to optimise yield and selectivity of single phenolic products of the hydrothermal lignin depolymerisation is of high interest.

D.2 Catalytic hydrothermal decomposition of catechol

Catechols are produced within the hydrothermal lignin degradation. Among the phenolic products they are the most abundant compound class with yields above 3 wt% (see Section 6.3). However, for the chemical industry catechols are less interesting as platform chemicals. In addition, they have a high potential to repolymerise under pyrolytic conditions [McM+04]. Thus the transformation of catechols into higher value products such as phenol or benzene deserves a more detailed observation. In Chapter5 the potential of the heterogeneous catalyst Raney-Ni for the conversion of catechol into phenol [For+12] was pointed out. This section shows a kinetic study of the hydrothermal conversion of catechol into phenol catalysed by Raney-Ni.

D.2.1 Results For the purpose of a detailed investigation of the effects of Raney-Ni on the conversion of catechol into phenol a set of 23 experiments (Exp. 103-125, see Appendix E.1) was carried out at temperatures between 473 K and 573 K and residence times between 30 and 180 min. Catechol was therefore diluted in desalinated water and together with Raney-Ni and elemental H2 treated in a 10 mL batch autoclave (MA2). In Figure D.3 the results of different experiments at 573 K and different residence times are demonstrated showing the mass fractions related to the input mass, that is the sum of the mass of

165 D.2. Catalytic hydrothermal decomposition of catechol

1 0 0

8 0 O t h e r s ) H

% 2 ( 6 0 C H n 4 o i

t C O 2 c

a C y c l o h e x a n o l

r 4 0 f

C y c l o h e x a n o n e s

s P h e n o l a 2 0 C a t e c h o l M

0 0 3 0 6 0 9 0 1 8 0 R e s i d e n c e t i m e ( m i n )

Figure D.3: Mass fractions of the input (residence time = 0) and output at 573 K and diﬀerent residence times after hydrothermal decomposition of catechol in a 10 mL-MA2; 5 ml input of aqueous solution (catechol/water = 10 g/L); 2.2 mg Raney-Ni catalyst

catechol and H2. Water was not considered to take part in any reaction. This is a simplification of the reaction system since water gas shift reaction and other reactions might occur and involve water. However, water is abundantly present throughout the reaction and thus changes of the water concentration may not influence the reaction kinetics. The results at 573 K show that phenol is produced in significant amounts as primary product. The amount of phenol decreases towards longer reaction times giving way to other phenol derivatives, such as cyclohexanol and cyclohexanone as well as to gaseous products such as CO2 and CH4. The elemental H2 is partially consumed within the ongoing reactions. In the following the reaction pathways of hydrothermal decomposition of catechol will be fathomed. Nimmanwudipong et al. showed that cyclohexanone is an important intermediate compound especially within the decomposition of methoxyphenols and is converted into cyclohexanol [Nim+11a; Nim+11b]. The rank of cyclohexanol and cyclohexanone could not be revealed by delplot-analysis for the present data. However, both compounds are produced in significant concentrations at high temperatures which indicates these substances to be secondary or tertiary products. The rank of the gaseous compounds as well stays unclear. Delplot analysis did not reveal any

166 Chapter D. Extended studies of catalysts

illuminating hints. However, CO2 and CH4 appear at temperatures above 523 K and increased residence times. Hence, they are probably secondary or tertiary products formed within the decomposition of phenol and/or cyclohexanol and cyclohexanone. Gasiﬁcation of catechol, that is the direct conversion of catechol into gaseous products, was not assumed. The focus of the kinetic studies lied on the formation and decomposition of phenol. Hence, phenol was assumed to be the only primary product derived from catechol.

OH O OH OH

k k 1 OH 2a H H 2 2 k CO 2b 2 CH4

Figure D.4: Reaction scheme of the catalytic decomposition of catechol in water at temperatures between 523 K and 573 K based on delplot-analysis and literature research.

The hydrothermal decomposition of catechols can thus be modelled with the reaction pathways displayed in Figure D.4. The conversion of catechol into phenol (k1) and the decomposition of phenol (k2a+k2b) as subsequent reaction is suggested to be sufficient to describe the concentration of phenol in the product mixture. The two reactions are treated like two first order reactions in series. The author is aware of the fact that the second reaction, that is the decomposition of phenol is not an intrinsic elementary reaction and might be a combination of two or more parallel reactions, e.g. reaction 2a and 2b. These assumptions lead to the following Equation D.1 for the calculation of the concentration of Catechol CA in dependence on the residence time τ considering the initial concentration of catechol CA,0 and the rate coefficient of the first order decomposition of catechol k1.

−k1τ CA = CA0 · e (D.1)

The concentration of phenol CP can thus be calculated considering k1

167 D.2. Catalytic hydrothermal decomposition of catechol

1 0 0 a ) 1 0 0 b ) ) L / l 2 2

o R = 0 . 9 7 6 R = 0 . 6 8 9 m

m 5 0 5 0 (

l o h c e t a 0 0 C 0 5 0 1 0 0 1 5 0 2 0 0 0 5 0 1 0 0 1 5 0 2 0 0 1 0 0 c ) 1 0 0 d ) ) L / l 2 2 o R = 0 . 9 4 5 R = 0 . 9 8 5 m

m 5 0 5 0 (

l o h c e t a 0 0 C 0 5 0 1 0 0 1 5 0 2 0 0 0 5 0 1 0 0 1 5 0 2 0 0 R e s i d e n c e t i m e ( m i n ) R e s i d e n c e t i m e ( m i n )

Figure D.5: Concentration of catechol in the product after hydrothermal decomposition of catechol depending on the residence time at a) 573 K, b) 533 K, c) 523 K, d) 513 K; Data points represent experimental values, continuous line calculated according to Equation D.1; 5 ml input of catechol/water = 10 g/L; 2.2 mg Raney-Ni catalyst.

and the rate coeﬃcient of the phenol decomposition k2.

k 1 −k1τ −k2τ CP = CA0 · (e − e ), k2 = k2a + k2b. (D.2) k2 − k1

The optimised rate coefficient of the decomposition of catechol k1 is determined by minimising the difference between the concentration CA calculated with Equation D.1 and the experimentally determined concentration CA. The curves which are described by Equation D.1 providing the best fit to the experimentally determined CA are plotted over the residence time at different temperatures (see Figure D.5). The resulting optimised rate coefficients k1 an k2 between 513 K and 573 K are plotted in an Arrhenius- diagram (see Figure D.7). Towards increasing temperatures the value of the

168 Chapter D. Extended studies of catalysts

8 0 8 0 a ) b ) )

L 2 /

l R = 0 . 9 4 3 o

m 4 0 4 0 m (

o 2 n R = 0 . 8 9 0 e h

P 0 0 0 5 0 1 0 0 1 5 0 2 0 0 0 5 0 1 0 0 1 5 0 2 0 0 8 0 8 0 2 c ) 2 d )

) R = 0 . 9 0 3 R = 0 . 8 6 0 L / l o

m 4 0 4 0 m (

l o n e h

P 0 0 0 5 0 1 0 0 1 5 0 2 0 0 0 5 0 1 0 0 1 5 0 2 0 0 R e s i d e n c e t i m e ( m i n ) R e s i d e n c e t i m e ( m i n )

Figure D.6: Concentration of phenol in the product after hydrothermal decomposition of catechol depending on the residence time at a) 513 K, b) 523 K, c) 533 K and d) 573 K; Data points from experiments, continuous line calculated according to Equation D.2; 5 ml input of catechol/water = 10 g/L; 2.2 mg Raney-Ni catalyst.

optimised k1 increases. The temperature dependence of a kinetic rate coefficient can be described by the Arrhenius-Equation C.3. Optimisation of the linear approach of k1 in the Arrhenius-diagram yields the kinetic parameters EA,j and Aj, which are shown in Table D.1. The experimentally determined concentration of catechol is plotted over the residence time at different temperatures in Figure D.5. The curves resulting from the optimisation of k1 are added. The regression of the catechol concentration was acceptable. For all temperatures except 523 K R2 was above 0.9. The optimisation of the rate coefficient of the decomposition of phenol k2 is carried out analogously. The difference between CP calculated by Equation D.2 and the experimentally determined concentration CP is minimised. The rate coefficient k1 and the initial concentration of catechol were hereby fixed parameters in Equa- tion D.2; k1, EA1 and Arrhenius factor A1 result from the prior optimisation.

169 D.2. Catalytic hydrothermal decomposition of catechol

- 3 . 0

- 3 . 5

- 4 . 0 ) ' k /

k - 4 . 5 ( l n ( k / k ' )

n 1 l - 5 . 0 l n ( k 2 / k ' ) - 5 . 5 0 . 0 0 1 7 0 . 0 0 1 8 0 . 0 0 1 9 0 . 0 0 2 0 T - 1 ( K - 1 )

Figure D.7: Arrhenius plot of pseudo-ﬁrst order rate coeﬃcients of the hydrothermal Raney-Ni catalysed decomposition of catechol (circular data points) and phenol (rectangular data points), k0 = 1min−1

Table D.1: Parameters describing the decomposition of catechol and phenol within hydrothermal catechol decomposition catalysed by Raney-Ni. rj kj EA,j Aj T = 513 K T = 523 K T = 533 K T = 573 K min−1 kJ/mol min−1 r1 0.017 0.024 0.029 0.031 23 ± 4 0.6 ± 0.4 r2 0.006 0.010 0.012 0.028 58 ± 5 3.7 ± 0.4

Figure D.6 shows the concentration of phenols. The scattered data points represent the experimentally determined concentrations, the curves show the calculated concentration resulting from the optimisation of k2. The rate co- eﬃcient of phenol decomposition k2 in dependence of the temperature is plotted in Figure D.7. The coeﬃcient of determination R2 was above 0.85 for all temperatures.

D.2.2 Discussion The experimental data shows that the concentration of catechol is declining with increasing temperature and residence time. The concentration of phenol increases at short residence times and decreases towards longer residence times in dependence on the temperature. The modelling of the experimental yields shows that a simple two step reaction pathway comprising the primary

170 Chapter D. Extended studies of catalysts conversion of catechol into phenol followed by the decomposition of phenol is able to describe the yields of catechol and phenol. The model fit is acceptable since in the temperature range from 513 to 573 K the coefficient of determination was above 0.65 and mostly above 0.85. The kinetics of the mentioned reaction series was analysed. The results show rather low apparent activation energies, about 23 kJ/mol for the decomposition catechol and 58 kJ/mol for the decomposition phenol. The activation energy calculated for the catechol decomposition within the hydrothermal lignin degradation (see Chapter 6.3) was 55 kJ/mol, the non-catalytic hydrothermal catechol decomposition studied by Wahyudiono, Sasaki, and Goto [WSG09] yielded an apparent activation energy of 51 kJ/mol, which is about two times higher than the apparent activation energy of the catalytic decomposition. This indicates that probably diffusion effects dominate the reaction rate of the catalytic catechol decomposition. These effects are especially important with increasing temperatures as the relatively low increase of reaction rate coefficient from 0.029 to 0.031 min−1 between 533 and 573 K shows (see Table D.1). However, it becomes clear that the reaction rate of hydrothermal catechol decomposition can be increased significantly by the application of Raney-Ni compared to experiments without catalyst. Wahyudiono et al. [WSG09] reported rate coefficients of about 0.0001 min−1 for non-catalytic catechol decomposition in supercritical water at 683 K, whereas catalytic catechol decomposition yields rate coefficients of about 0.03 min−1 at 533 K.

171 D.2. Catalytic hydrothermal decomposition of catechol

172 Appendix E

Experimental results

E.1 Tables of experimental results

The following tables contain the data of the experiments carried out. Listed are

• the type of reactor,MA meaning microbatch-autoclave, CSTR meaning continuous stirred tank reactor. Two types ofMA were used that areMA1 which is the simple type andMA2 which is the gas-input version with the possibility of pressurising the reactor with any kind of gas,

• the reaction volume VR which is equal to the volume of the reactor, • the (mean) residence time τ (τ 0), • the reaction temperature T , • type and initial input mass of the Biomass (Lignin), • type and initial input mass of the solvent,

• type and initial input mass of the added gas or H2-donor, • type and initial input amount of the catalyst. The initial amount for homogeneous catalysts is given in mol/L, for heterogeneous catalysts in mg, • the analysis methods applied on the product phases of each experiment, that are extraction (Ext),GC, KFT, UV-Vis, FT-IR,

173 E.1. Tables of experimental results

• the mass of the liquid product phase which was gravimetrically determined, • the output mass of gaseous and solid phase determined by the sum of the mass of each constituent and by gravimetric analysis respectively,

• the loss for each experiment which is the diﬀerence between the sum of the mass of all input materials and the sum of the mass of all output P P materials ( mi,0 − mi) • the yield of the gaseous compounds that are methane, ethene, ethane, propene, propane, butane, carbon dioxide, hydrogen, carbon monoxide and other gases, • the recovered amount of water and ethanol related to the input mass of lignin, • the yield of all liquid products detected byGC-analysis, that are phenol (Phe), 2-methylphenol (MePhe), 4-MePhe, 2-EtPhe, 4-EtPhe, guaiacol (Gua), 4-methylguaiacol (MeGua), 4-ethylguaiacol (EtGua), syringol (Syr), catechol (CatOH), 4-methylcatechol (MeCatOH), 4- ethylcatechol (EtCatOH), cyclohexanol (CyclohexOH) and cyclohexanone (CyclohexO).

All yields are given in mg/gBiomass. If the displayed value of the yield is zero, the detected yield was smaller than 0.5 mg/gBiomass. If the compound was not detected even in small quantities, this is indicated by not detected (n.d.). If the compound was not quantiﬁed analytically this is indicated by not available (n.a.). For experiments in the CSTR the input and output amounts are given in g/h.

174 Chapter E. Experimental results c a 30 Ext . GC 17 Ext . GC 10 Ext . GC 30 Ext . GC 22 Ext . GC 32 Ext . GC 30 Ext . GC b b b b 3 3 3 3 O O O O 2 2 2 2 b 2 2 Type Amount 0 3 HCl 1.81 . GC Ext 4 HCl 0.33 . GC Ext 3 Raney-Ni 234 Ext . GC 14 Raney-Ni 234 Ext . GC 13 Rh(5 wt%)/C 14 Co 12 Ext . GC 11 Pt(5 wt%)/Al 14 Pt(5 wt%)/Al 14 PdCl 1114 Raney-Ni Raney-Ni3862 59 Raney-Ni 23 Raney-NiExt . GC Ext . GC 234 234 Ext . GC Ext . GC i, 155 Co155 Pt(5 wt%)/Al 27 Ext . GC 188 Pt(5 wt%)/Al 155 PdCl 155188 Raney-Ni155 Raney-Ni188 Raney-Ni155 Raney-Ni Raney-Ni 586 1172 Ext . GC 117 Ext . GC 352 Ext . GC 234 Ext . GC Ext . GC 155155 HCl188 Acetic Acid188 Acetic Acid155 NaOH188 None155 None Raney-Ni 2.92 0.58 . GC Ext 1.81 Ext . GC . GC Ext 0.12 . GC Ext 234 0 Ext . GC 0 Ext . GC Ext . GC m 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Type 0 i, m OO 3.0 3.0 H2 N 14 KOH 0.12 . GC Ext O 3.0 N OOO 3.0O 3.0 N O 3.0 N O 3.0 N O 3.0 N O 3.0 N 2.8 N 2.8 N H O 3.0 H OOO 3.0 3.0 N 3.0 H N O 3.0 N O 3.0 H O 3.0 H O 3.0 N O 3.0 H OOO 3.0O 3.0 H O 2.5 H O 2.0 N O 2.9 N O 2.7 N O 2.8 N O 2.8 N 2.8 N 2.8 N N 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Type 0 i, m Biomass Solvent H-Donor/Gas Catalyst Analysis Type T ) 0 τ ( τ R V Type Table E.1: Experimental data: reaction conditions, input parameters, employed analytical methods. Exp. Reactor 1 MA 22 MA 5.5 5.5 30 30 633 633 spruce- Lig spruce- Lig 0.400.40 H H #3 MA 2 5.5 ml 30 min 633 spruce- Lig K0.40 H g g mg 4 MA 2 5.5 30 633 spruce- Lig 0.40 H 5 MA 26 MA 2 5.57 MA 2 5.58 MA 2 5.59 MA 2 5.5 3010 MA 2 5.5 3011 MA 2 633 5.5 30 633 5.5 spruce- Lig 30 633 spruce- Lig 30 6330.40 spruce- Lig 30 6330.40 spruce- Lig 30 H 6330.40 spruce- Lig H 6330.40 spruce- Lig H 0.40 spruce- Lig H 0.40 H 0.40 H H 12 MA 2 5.5 30 633 spruce- Lig 0.40 H 13 MA 2 5.5 30 633 spruce- Lig 0.40 H 14 MA 215 MA 2 5.5 5.5 30 30 633 633 spruce- Lig spruce- Lig 0.400.40 H H 16 MA 2 5.5 30 633 spruce- Lig 0.40 H 17 MA 2 5.5 30 633 spruce- Lig 0.40 H 18 MA 2 5.5 30 633 spruce- Lig 0.40 H 19 MA 2 5.5 30 633 spruce- Lig 0.40 H 20 MA 2 5.5 30 633 spruce- Lig 0.40 H 21 MA 2 5.5 30 633 spruce- Lig 0.40 H 22 MA 223 MA 2 5.524 MA 2 5.525 MA 2 5.526 MA 2 5.5 3027 MA 2 5.5 3028 MA 2 633 5.5 3029 MA 2 633 5.5 spruce- Lig 3030 MA 2 633 5.5 spruce- Lig 30 6330.40 5.5 spruce- Lig 30 6330.40 spruce- Lig 30 H 6330.40 spruce- Lig 30 H 6330.40 spruce- Lig 30 H 6330.40 spruce- Lig H 6330.40 spruce- Lig H 0.40 spruce- Lig H 0.40 H 0.40 H H

175 E.1. Tables of experimental results c a Type Amount 0 11 Al-Ni 57 . GC Ext i, 2.0 Raney-Ni 67.2 Ext . GC 2.0 Raney-Ni 67.2 Ext . GC 2.0 Raney-Ni 67.2 . GC Ext 2.0 Raney-Ni 67.2 . GC Ext 2.0 Raney-Ni 67.2 . GC Ext 2.0 Raney-Ni 67.2 . GC Ext 2.0 Raney-Ni 67.2 . GC Ext 2.0 Raney-Ni 67.2 . GC Ext 191 Al-Ni 53 . GC Ext 132 Raney-Ni 234 Ext . GC m 2 2 2 2 2 2 2 2 2 2 2 Type 0 i, m O 5.0 H O 5.0 H O 5.0 H O 5.0 H O 5.0 H O 5.0 H O 5.0 H O 5.0 H O 3.0 N OOO 2.8 3.0 CO 3.0 CO H 56 70 Raney-Ni None 234 . GC Ext 0 Ext . GC O 2.8 N 2 2 2 2 2 2 2 2 2 2 2 2 2 Type 0 i, m Biomass Solvent H-Donor/Gas Catalyst Analysis Type T ) 0 τ ( τ R V Type 44 MA 245 MA 246 MA 2 5.547 MA 2 5.548 MA 2 5.549 MA 2 5.5 1550 MA 2 5.5 3051 MA 2 633 5.5 4552 MA 2 633 5.5 60 wheat- Lig 53 MA 2 633 5.5 90 wheat- Lig 54 MA 1 633 1200.33 5.5 wheat- Lig 55 MA 1 633 2400.33 5.5 wheat- Lig 63356 MA 1EtOH 3600.33 5.0 wheat- Lig 63357 MA 1EtOH wheat- Lig 480 0.33 5.0 633 118058 MA 1EtOH 10.0 wheat- Lig 2.00.33 63359 MA 1EtOH 0.33 10.0 wheat- Lig 660 6332.0 FA60 MA 1EtOH 0.33 10.0 wheat- Lig 6602.0EtOH 653 150 wheat- Lig FA61 MA 10.33 10.02.0EtOH 653 150 FA62 0.33 10.0 268.4 None 6532.0EtOH 0.33 FA2.063 10.0 268.4 None 60 653EtOH None wheat- Lig FA2.064 EtOH 268.4 60 FACSTR None 653 140 wheat- Lig 2.065 268.4 FACSTR 0.35 None 653 1402.0 268.4 wheat- Lig 0.00 653 FA2.0200CSTR 0.35 None 268.4EtOH wheat- Lig 0.00 653 FA200CSTR None 268.4EtOH wheat- Lig FA0.35 EtOH None200 268.4EtOH wheat- Lig 0.35 None3.2200 51EtOH 0.35 268.41.6 0 GC . FT-IR None 268.43.2 59EtOH 0.35 FA1.6 653 0 GC . FT-IR . KFT EtOH None None 39 None3.2 FA 653GC . FT-IR 0 EtOH FA 56 wheat- Lig 3.2 653 0 GC . FT-IR . KFT 390.4 FA3.2 wheat- Lig 633 0 GC . FT-IR 390.4 FA3.2 6.22 wheat- Lig 0 GC . FT-IR None 0 152.5 FA6.33 wheat- Lig 0 GC . FT-IR KFT 390.4 None FA EtOH None None6.46 0 GC . FT-IR 390.4EtOH None 390.46.37 0 GC . FT-IR KFT 59.0 EtOH 0 GC . FT-IR KFT None390.460.0 NoneEtOH FA 61.3 None FA 60.4 FA 0 GC . FT-IR KFT FA 0 GC . FT-IR KFT 7.6 0 GC 0 GC 7.7 0 GC . FT-IR KFT None7.9 0 GC . FT-IR KFT None 0 GC . FT-IR KFT 7.8 None 0 GC . FT-IR KFT None 0 GC . FT-IR KFT 0 GC . FT-IR KFT 0 GC . FT-IR KFT 0 GC . FT-IR KFT 43 MA 2 10 60 623 Guaiacol 0.05 H 42 MA 2 10 60 603 Guaiacol 0.05 H 41 MA 2 10 60 573 Guaiacol 0.05 H 40 MA 2 10 60 523 Guaiacol 0.05 H 39 MA 2 10 60 623 Catechol 0.05 H 38 MA 2 10 60 603 Catechol 0.05 H 37 MA 2 10 60 573 Catechol 0.05 H 36 MA 2 10 60 523 Catechol 0.05 H 35 MA 2 5.5 30 633 spruce- Lig 0.40 H 33 MA 234 MA 2 5.5 5.5 30 30 633 633 spruce- Lig spruce- Lig 0.400.40 H H 32 MA 2 5.5 30 633 spruce- Lig 0.40 H Exp. Reactor 31 MA 2 5.5 30 633 spruce- Lig 0.40 H # ml min K g g mg

176 Chapter E. Experimental results c a Type Amount d 0 i, m Type d 0 i, m OOOO 2.5O 2.5 NoneO 2.5 NoneO 2.5 NoneO 2.5 None 0O 2.5 None 0O 2.5 None None 0O 2.5 None None 0O 2.5 None None 0O 2.5 None None 0O 2.5 None None 0 0 O 2.5 None None 0 0 O 2.5 NoneExt . GC FT-IR None 0 0 O 2.5 NoneExt . GC FT-IR None 0 0 O 2.5 NoneExt . GC FT-IR None 0 0 O 2.5 NoneExt . GC FT-IR None 0 0 O 2.5 NoneExt . GC FT-IR None 0 0 O 2.5 NoneExt . GC FT-IR None 0 0 O 2.5 NoneExt . GC FT-IR None 0 0 2.5 NoneExt . GC FT-IR None 0 0 2.5 NoneExt . GC FT-IR None 0 0 2.5 NoneExt . GC FT-IR None 0 0 NoneExt . GC FT-IR None 0 0 NoneExt . GC FT-IR 0 0 NoneExt . GC FT-IR 0 0 NoneExt . GC FT-IR 0 0 NoneExt . GC FT-IR 0 NoneExt . GC FT-IR 0 Ext . GC FT-IR 0 Ext . GC FT-IR 0 . GC FT-IR Ext 0 . GC FT-IR Ext 0 . GC FT-IR Ext . GC FT-IR Ext 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Type d 0 i, m BiomassType Solvent H-Donor/Gas Catalyst Analysis T ) 0 τ ( τ R V Type Exp. Reactor 66 67 68 CSTR 69 MA 2CSTR 200CSTR 200 5.5200 44 37 15 673 21 653 593 wheat-Lig 653 wheat- Lig spruce- Lig wheat-Lig 6.52 8.77 0.33 EtOH 15.43 EtOH H EtOH 61.8 83.1 146.3 FA FA FA 8.010.618.7 None None None 0 GC . FT-IR KFT 0 GC . FT-IR KFT 0 GC . FT-IR KFT 70 MA 271 MA 272 MA 2 5.573 MA 2 5.574 MA 2 5.575 MA 2 5.5 3076 MA 2 5.5 45 59377 MA 2 5.5 60 59378 MA 2 5.5 spruce- Lig 90 593 12079 MA 2 5.5 spruce- Lig 593 24080 MA 2 5.50.33 spruce- Lig 593 48081 MA 2 5.50.33 spruce- Lig 593 H spruce- Lig 82 MA 2 5.50.33 15 593 H spruce- Lig 83 MA 2 5.50.33 30 H 0.33 613 spruce- Lig 84 MA 2 5.5 45 H 0.33 61385 MA 2 5.5 H spruce- Lig 600.33 61386 MA 2 5.5 H spruce- Lig 90 613 12087 MA 2 5.50.33 H spruce- Lig 613 24088 MA 2 5.50.33 spruce- Lig 613 H 48089 MA 2 5.50.33 spruce- Lig 613 H spruce- Lig 90 MA 2 5.50.33 15 613 H spruce- Lig 5.50.33 30 H 0.33 633 spruce- Lig 5.5 45 H 0.33 633 H spruce- Lig 600.33 633 H spruce- Lig 90 633 1200.33 H spruce- Lig 6330.33 spruce- Lig 633 H 0.33 spruce- Lig H spruce- Lig 0.33 H 0.33 H 0.33 H H # ml min K g g mg

177 E.1. Tables of experimental results c a Type Amount 0 44 Raney-Ni4 Raney-Ni4 Raney-Ni4 Raney-Ni4 Raney-Ni 22 4 Raney-Ni 22 4. GC Ext Raney-Ni 22 4. GC Ext Raney-Ni 22 4. GC Ext Raney-Ni 22 4. GC Ext Raney-Ni 22 4. GC Ext Raney-Ni 22 4Ext . GC Raney-Ni 22 4Ext . GC Raney-Ni 22 4Ext . GC Raney-Ni 22 4Ext . GC Raney-Ni 22 4Ext . GC Raney-Ni 22 4Ext . GC Raney-Ni 22 Ext . GC Raney-Ni 22 Ext . GC 22 Ext . GC 22 Ext . GC 22 Ext . GC 22 Ext . GC Ext . GC i, m 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Type 0 i, m OOO 2.5O 2.5 NoneO 2.5 NoneO 2.5 NoneO 2.5 NoneO 0 2.5 NoneO 0 2.5 None NoneO 0 2.5 None NoneO 0 2.5 None NoneO 0 2.5 None NoneO 0 2.5 None NoneO 0 2.5 None NoneO 0 5.0 None NoneO 0 0 5.0 None H O 0 0 5.0Ext . GC FT-IR None H O 0 0 5.0Ext . GC FT-IR None H O 0 0 5.0Ext . GC FT-IR None H O 0 5.0Ext . GC FT-IR None H O 0 5.0Ext . GC FT-IR H O 0 5.0Ext . GC FT-IR H O 0 5.0Ext . GC FT-IR H O 0 5.0Ext . GC FT-IR H O 0 5.0Ext . GC FT-IR H O 0 5.0Ext . GC FT-IR H O 0 5.0Ext . GC FT-IR H O 5.0Ext . GC FT-IR H O 5.0 H O 5.0 H 5.0 H 5.0 H H 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Type 0 i, m Type Biomass Solvent H-Donor/Gas Catalyst Analysis T ) 0 τ ( τ R V Type Exp. Reactor 91 MA 292 MA 2 5.593 MA 2 5.594 MA 2 5.595 MA 2 240 5.596 MA 2 360 5.5 63397 MA 2 480 5.5 63398 MA 2 spruce- Lig 5.5 633 1599 MA 2 spruce- Lig 5.5 300.33100 MA 2 653 spruce- Lig 5.5 450.33101 MA 2 653 H 5.5 spruce- Lig 600.33102 MA 2 653 H 5.5 spruce- Lig 90103 MA 2 653 1200.33 H 5.5 spruce- Lig 104 MA 2 653 2400.33 spruce- Lig 653 10 H 105 MA 2 3600.33 spruce- Lig 653 10 H 106 MA 2 spruce- Lig 4800.33 653 10 H 107 MA 2 spruce- Lig 0.33 653 10 H 0.33 30108 MA 2 spruce- Lig 10 H 0.33 90109 MA 2 spruce- Lig H 473 180 100.33110 MA 2 H 473 100.33 Catechol 473 60111 MA 2 H 10 Catechol 90112 MA 2 H 493 Catechol 10 60113 MA 2 0.05 493 10 Catechol 90114 MA 2 0.05 503 180 H 0.05 10 Catechol115 MA 2 503 H 10 Catechol 503 60116 MA 2 0.05 H 10 Catechol 90117 MA 2 0.05 513 Catechol 180 H 10118 MA 2 0.05 513 H 10 Catechol 513 30119 MA 2 0.05 H 0.05 10 Catechol 90120 MA 2 523 Catechol H 10 90 0.05 H 523 180 10 Catechol 0.05 523 180 H 0.05 Catechol 523 H Catechol 523 60 0.05 H Catechol 90 0.05 533 Catechol H 0.05 533 H 0.05 Catechol H 0.05 Catechol H 0.05 H 0.05 H H # ml min K g g mg

178 Chapter E. Experimental results c a Type Amount 0 44 Raney-Ni4 Raney-Ni4 Raney-Ni4 Raney-Ni Raney-Ni 22 22 Ext . GC 22 Ext . GC 22 Ext . GC 22 Ext . GC Ext . GC i, 2.82.8 Pd(10 %wt)/C2.8 Pd(10 %wt)/C2.8 Pd(10 %wt)/C2.8 Pd(10 %wt)/C2.8 Pd(10 9.0 GC %wt)/C2.8 Pd(10 8.4 GC %wt)/C2.8 Pd(10 8.0 GC %wt)/C2.8 Pd(10 8.4 GC %wt)/C2.8 Pd(10 8.6 GC %wt)/C2.8 Pd(10 8.8 GC %wt)/C2.8 Pd(10 8.2 GC %wt)/C2.8 Pd(10 8.1 GC %wt)/C2.8 Pd(10 8.2 %wt)/C2.8 Pd(10 7.7 %wt)/C2.8Ext . GC Pd(10 7.9 %wt)/CExt . GC Pd(10 8.8 %wt)/CExt . GC 8.5 Ext . GC 8.9 Ext . GC 8.1 Ext . GC 8.7 Ext . GC Ext . GC m 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Type 0 i, m OOO 5.0O 5.0O H 5.0 H 5.0 H 5.0 H H OOO 2.0O 2.0O H 2.0O H 2.0O H 2.0O H 2.0O H 2.0 H 2.0O H 2.5 H O FA 2.5 OO FA 268.4 2.5O 2.5O None None268.4 2.5 None 2.5 None None 2.5 None 0 FA 0 None 0 None 0268.4 None None None 0 Ext . GC FT-IR 0 Ext . GC FT-IR 0 0 . GC FT-IR Ext 0 . GC FT-IR Ext 0 0 . GC FT-IR Ext . GC FT-IR Ext Ext . GC FT-IR 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Type 0 i, m Type Biomass Solvent H-Donor/Gas Catalyst Analysis T ) 0 τ ( τ R V Type Exp. Reactor #121 MA 2122 MA 2 10123 MA 2 10124 MA 2 10125 MA 2 180 10126 MA 2 ml 10 533 30127 MA 2 5.5 60128 MA 2 573 Catechol 5.5 min 90129 MA 2 573 180 5.5 Catechol130 MA 2 573 5.5 60 K 0.05 Catechol 573131 MA 2 5.5 60 Catechol132 MA 2 573 0.05 H Catechol 5.5 60133 MA 2 593 0.05 5.5 spruce- Lig 60 H 134 MA 2 613 0.05 5.5 spruce- Lig 60 H 0.05135 MA 2 6330.16 5.5 spruce- Lig 60 H 136 MA 2 653 H 0.15 5.5 spruce- Lig 60EtOH 137 MA 2 6730.16 5.5 spruce- Lig 60EtOH 138 MA 2 6930.16 5.5 spruce- Lig 60EtOH 139 MA 22.0 7130.16 5.5 spruce- Lig 60 gEtOH 140 MA 22.0 5730.16 H 5.5 spruce- Lig 60EtOH 141 MA 22.0 5930.16 H 5.5 spruce- Lig 60EtOH 142 MA 22.0 6130.16 H 5.5 spruce- Lig 60EtOH 143 MA 22.0 6330.16 H 5.5 spruce- Lig 60 EtOH 144 MA 22.0 6530.16 H 5.5 spruce- Lig 60 H 145 MA 22.0 6730.17 H 5.5 spruce- Lig 60 H 146 MA 22.0 6930.16 H 5.5 spruce- Lig 60 H g147 MA 2 7130.16 5.5 H spruce- Lig 60 H 148 MA 2 633 2400.15 5.5 spruce- Lig H 149 MA 2 633 2400.16 5.5 spruce- Lig 633 1180 H 150 MA 20.16 5.5 spruce- Lig 633 1180 H spruce- Lig 5930.33 5.5 1180 H spruce- Lig 6130.33 spruce- Lig 1180 H 0.33 mg 633 spruce- Lig 1180EtOH 0.33 653 H 0.33 spruce- Lig 633EtOH 0.33 spruce- Lig H 2.0 0.33 spruce- Lig H 0.33 FA 2.0 H 0.33 H FA H 268.4 None268.4 None 0 GC . FT-IR GC . FT-IR 0

179 E.1. Tables of experimental results c Type Amount 0 i, m Type 0 i, m in case of homogeneous catalysts. Type mol/L 0 . i, m g/h Type Biomass Solvent H-Donor/Gas Catalyst Analysis T ) 0 τ ( τ in case of heterogeneous catalysts, mg R V Type Catalyst/Carrier material Input unit of all input streams in case of CSTR is Unit of amount: GC =Gas Chromatography, FT-IR =Infra red spectroscopy, Ext =Extraction, KFT =Karl-Fischer-titration. a b c d Exp. Reactor #151 MA 2152 MA 2 5.5153 MA 2 5.5154 MA 2 5.5 1180155 MA 2 5.5 1180156 MA 2 633 ml 5.5157 MA 2 633 5.5 15 spruce- Lig 158 MA 2 5.5 min 30 spruce- Lig 159 MA 2 6330.33 5.5 45160 MA 2 6330.33 5.5 60 spruce- Lig KEtOH 161 MA 2 633 5.5 90 spruce- Lig EtOH 633 1200.33 5.5 spruce- Lig 2.0 633 2400.33 spruce- Lig 633EtOH 2.0 3600.33 FA spruce- Lig 633EtOH spruce- Lig 4800.33 None 633EtOH 2.0 spruce- Lig 0.33 633EtOH 0.33 268.42.0 spruce- Lig NoneEtOH 0.33 2.0 spruce- Lig EtOH None None 00.33 2.0 EtOH None g0.33 2.0 NoneEtOH None2.0 0 EtOH None2.0 0 None None2.0 0GC . FT-IR 0 None None2.0 0 None None 0 0 None None 0 None 0GC . FT-IR None 0 GC . FT-IR g 0 None 0 GC . FT-IR 0 NoneGC . FT-IR 0 None 0 GC . FT-IR 0 GC . FT-IR 0 GC . FT-IR 0 GC . FT-IR mg 0 GC . FT-IR 0 GC . FT-IR

180 Chapter E. Experimental results O EtOH 2 O and . EtOH 2 CO Others H 2 H 2 CO 10 H 4 C Biomass 8 H 3 C 6 H 3 C 6 H 2 C 4 H 2 C 4 CH a wt% mg/g Biomass Table E.2: Experimental data: Results of bulk components, gaseous compounds, H Exp. #Liq.1 Char23 g Gas4 2.7 mg/g 5 Loss 2.36 1.8 2837 2.5 6388 2.6 441 5969 2.5 612 29510 2.9 455 15311 10 2.4 473 78212 23 2.6 181 395 2.413 40 432 298 2.614 3 16 457 25 502 2.615 15 309 2.216 0 2 18 51 288 2.517 0 344 0 8 2.018 398 0 4 21 726 1.919 347 3 16 833 323 0 2.120 423 0 426 2 18 2.421 0 411 4 363 0 0 10 17 2.822 413 3 372 0 25 2.823 0 436 37 212 3 17 2.624 357 0 38 383 1 0 3 31 2.625 0 1 258 313 0 3 36 2.226 0 256 364 0 0 2 29 1.827 0 212 248 1 0 3 21 2.4 028 1 0 477 0 1 3 10 2.129 0 0 1 377 1 3 11 2.2 030 4 0 101 1 2 3 14 1015 2.3 1 4 0 1 1 4 17 1 2.8 0 0 0 6 952 605 1 1 2 2.4 0 0 0 13 19 0 0 576 1 19 0 0 240 267 0 0 1 7 26 19 0 0 0 0 742 16 1 47 117 685 1 0 12 1 0 0 443 1 0 129 28 0 3 1 21 464 15 1 387 23 3 0 12 3 1 0 0 511 1 0 287 22 1 1 0 0 1 6 10 0 29 6 0 0 1 0 41 497 0 0 302 19 2n.a. n.a. 0 0 27 0 1 278 2 0 18 1 0 1 0 4n.a. n.a. 680n.a. n.a. 0 0 0 24 1 1 6 0 6 0 775 0 2 288 1 1 0n.a. n.a. 0 0 1n.a. n.a. 418 0n.a. 0 1 12 0 24 0 325 0n.a. 0 6 1 0 1 0 4n.a. n.a. n.a. 364 1 0 0 0n.a. 0 28 3 n.a. 202 1 n.a. 1 4 2 356 0 n.a. 2 0n.a. n.a. 2 0 3n.a. 273 0n.a. 4 5 0 0n.a. 1n.a. 2 20n.a. 353 n.a. 0 2 32n.a. n.a. 200 1 0n.a. n.a. 445n.a. 5 n.a. 1n.a. 1 2 0 0n.a. 3 32 334 2 3 14n.a. n.a. 1n.a. 2n.a. n.a. 961n.a. n.a. 1 6 13 27 811n.a. n.a. 579 2 1 n.a. 0n.a. 0n.a. n.a. 2n.a. 5 0 2n.a. n.a. 0n.a. n.a. 259 692 1n.a. 0 n.a. n.a. n.a. n.a. 408 2 0 n.a. 0 0n.a. n.a. 441 0 480n.a. 0n.a. n.a. n.a. n.a. 0 0n.a. n.a. 0 n.a. 0 0 n.a. n.a. n.a. 1 n.a. n.a. n.a. 0 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.

181 E.1. Tables of experimental results O EtOH 2 n.a. n.a. n.a. n.a. n.a. n.a. b b 8 15 CO Others H b b 2 3 4 H b b 2 6 235 CO b b 10 13 51 H 4 C Biomass b b 8 4 5 H 3 C b b 6 8 10 H 3 C b b 6 4 55 H 2 C b b 4 37 23 H 2 C b b 4 CH a -9 3 18 7 wt% mg/g b b Biomass 565758 3.359 3.160 3.0 91 3.2 122 3.2 159 2310 2260 174 1742 208 1757 -2 2 2216 7 31 3 35 1 21 21 48 50 38 86 218 96 246 52 90 18 91 17 217 4 30 35 3 16 10 56 30 51 9 1445 1311 11 0 44 1140 0 9 1319 65 1137 0 81 0 159 0 400 103 434 163 221 0 397 0 229 483 503 496 6010 0 0 5802 6330 55 1.1 n.d. 445 Exp. #Liq.31 Char3233 g Gas 2.534 2.4 mg/g 35 Loss 2.436 2.537 0 2.138 0 403 4.839 415 591 4.340 411 530 308 4.641 335 18 4.842 0 310 20 23 4.943 0 17 4.244 0 30 29 84 4.845 0 19 185 2 4.746 0 359 4 1.947 0 31 753 0 4 39 1.948 0 165 0 34 1.949 0 341 0 171 16 124 31 2.050 138 0 487 219 30 2.051 3 188 0 763 376 457 40 1.852 2 157 876 0 31 1033 1.853 0 0 71 1 95 113 32 30 1240 1.954 154 0 0 1 346 27 1.6 0 26 1091 0 80 507 1414 1.6 0 102 20 0 0 0 0 1 0.9 147 0 1985 0 1 23 1725 2 1 149 0 3 24 0 1961 1n.d. 0 6 0 19 3043 0 0 0 3 21 0 0 19 3 11 0 26 27130 0 0 1 0 24 17 0 8 1 9 0 0 0 0 0 22 21 18 554 0 6 15 0 0 22 472 0 26 125 0 0 20 0 0 20 302 25 82 23 0 0 0 0 297 0 0 0 0 0 29 23 134 0 0 0 0 151 35 180 204 0 0 0 68 61 0 4 444 140 0 22 4 0 8 12 296 0n.a. 39 20n.a. n.a. 0 7 0 33 0 122 0 174 37n.a. 94 0n.a. 199 0 n.a. n.a. 58 224 140 0 0 303 0 0 0n.a. 659 0 n.a. 256 0 0n.a. 0 0n.a. n.a. 347 918 5 0n.a. 795 0 1104 16 29n.a. n.a. 0n.a. 1164 46 245 n.a. 0 1596 0 0 0 987 1272 n.a. 0 0 0n.a. n.a. 0 n.a. 0 1328 1470n.a. 0 0 0 26 49 0n.a. 0 n.a. 46 0 42 0n.a. 0 0n.a. 5 n.a. 0 n.a. 16 0 n.a. 0 n.a. n.a. 0 n.a. 0 n.a. n.a. 5 0 n.a. 0 n.a. n.a. n.a. 233 n.a. 233 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 377 n.a. n.a. n.a. 515 880 n.a. n.a. n.a. n.a. n.a. n.a.

182 Chapter E. Experimental results O EtOH 2 CO Others H 2 H 2 CO 10 H 4 C Biomass 8 H 3 C 6 H 3 C 6 H 2 C 4 H 2 C 4 CH a wt% mg/g Biomass Exp. Liq.6162 Char6364 Gas 3.265 73.0 Loss 66 63.8 19167 68.6 7968 67.9 2271 4969 66.7 1092 1870 93.7 167.1 62 87971 3 -10 17 82872 2.4 78773 9 10 1142 2.5 39 6 2474 2.5 497 275 998 800 2.5 565 276 19 3 2.5 486 54 3677 18 40 2.4 48178 0 0 61 2.3 328 7 4079 222 30 84 2.0 486 76 1680 33 67 2.3 471 15 12 1381 90 2.5 386 17 35 1582 16 46 98 2.4 554 0 14 10583 41 0 2.5 493 22 36 0 17 14184 2.2 19 444 73 1 1785 31 19 75 2.4 0 486 186 0 31 93 2.1 0 28 444 23 0 187 0 95 2.4 429 1 1888 0 0 0 2 46 93 2.3 375 14 11389 0 0 3 2.3 0 366 18 1352 10690 0 0 0 0 2.4 0 0 498 0 15 152 0 25 2.4 0 0 0 476 0 1 181 0 16 2.2 512 0 408 0 1 0 0 26 67 2.4 0 0 106 0 337 1 0 123 0 0 10 2 16 0 322 400 122 0 0 0 3 0 0 336 241 19 436 0 0 0 5 19 56 0 407 151 0 8 0 6 18 331 0 446 0 0 0 0 166 0 7 0 677 0 274 0 193 0 1 17 11 2 24 441 411 0 0 6 265 0 0 4 18 0 0 331 5925 0 559 0 0 296 10 0 0 0 102 8 2 0 0 0 0 5 0 329 0 9364 38 0 28 0 278 0 5 0 463 206 0 55 111 0 0 0 0 0 78 431 0 0 0 0 153 507 0 8116 0 0 61 0 425 0 0 0 33 0 0 83 0 0 8122 0 0 416 0 2 8129 91 101 0 0 8374 0 6 0 136 0 0 0 0 0 0 0 6 8586 0 0 0 0 0n.a. 0 5 69 0 0 0n.a. 6 8755 84 0 0n.a. n.a. 0 0 6 2 86 0 0 0n.a. n.a. 0 2 82 105 0 0 0 n.a. n.a. n.a. 100 0 0n.a. n.a. n.a. 0 5 n.a. 145 0 0 0n.a. 0n.a. 8 n.a. 175 0 0 0n.a. n.a. 7 n.a. 0 10 0 n.a. n.a. 6 n.a. 58 110 0 0n.a. 4 n.a. n.a. 109 0 n.a. n.a. 2 n.a. 0n.a. 0 0n.a. n.a. 1 51 141 0 n.a. n.a. n.a. n.a. 157 10 n.a. n.a. 8 n.a. 0 0 n.a. n.a. 9 n.a. 0n.a. n.a. n.a. n.a. n.a. n.a. 3 5 n.a. n.a. 3 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. # g mg/g

183 E.1. Tables of experimental results O EtOH 2 CO Others H 2 H 2 CO 10 H 4 C Biomass 8 H 3 C 6 H 3 C 6 H 2 C 4 H 2 C 4 CH a wt% mg/g Biomass Exp. #Liq.91 Char9293 g Gas 2.394 2.4 mg/g 95 Loss 2.496 357 2.597 354 2.298 361 160 2.599 520 176 1.9100 345 165 21 2.0101 364 17 83 2.3102 339 161 2.0 17103 374 152 2.3 7104 344 15 165 2.3 359 7 24105 n.a. 171 362 6 16106 n.a. 180 351 33 203107 n.a. 0 1 6 29 207108 n.a. 003800000 00 5 21 240109 n.a. 003600000 00 31 7110 n.a. 0050000000 20 0 0 8111 92n.a. 003900000 0 19 0 8112 28n.a. 003200000 0 7 0 12113 63n.a. 002300000 0 10 8114 0 0n.a. 0 0 8115 0n.a. 0 21 0 0116 0n.a. 00800000 0 0 0117 n.a. 00 0 0 0 0118 0n.a. 0035000000 0 0 0119 0 0n.a. 007600000 0 0120 0 72n.a. 0 0 134 0 0 0n.a. 0 0 0 0 0n.a. 0 0 11 0 33 18 00 148 0 0 0 00 169 0 0 0 00 159 0 1 14 0 12 0 120 0 0 0 0 96 0 72 141 0 0 0 11 6 47 0 140 0 0 0 0 0 0 154 0 32 0 0 0 0 162 0 14 32 0 0n.a. 171 0 15 0 0 195n.a. 0 23 9 0 0 0 196 0n.a. n.a. 7 231 0 0 n.a. 0 4 0 0000n.a. 0 0n.a. n.a. 2 n.a. 0 0000 0 0 0 0 0n.a. 0 1 n.a. n.a. 0000 0 n.a. n.a. 0 0 n.a. 0 0000 0 n.a. n.a. 0 0000 7 12n.a. 0n.a. 0n.a. 0 0n.a. 92 0000n.a. n.a. n.a. n.a. 21 51 0 0n.a. n.a. n.a. n.a. 0 n.a. 0n.a. 0 0 0 n.a. 0000 0 0 n.a. n.a. 0 0n.a. 0000n.a. n.a. n.a. 0000 22 56n.a. n.a. 0n.a. 0 0 n.a. n.a. 0 n.a. 0 49 64n.a. 0 0 n.a. n.a. 0 n.a. 00 n.a. n.a. 9 0 0 0 0 n.a. n.a. 35 n.a. 0 0 80 n.a. 72 0n.a. n.a. n.a. 0 5 74n.a. 0 n.a. 0 0 n.a. 0 n.a. 41n.a. n.a. 0 n.a. 0 n.a. n.a. 0 0 n.a. 0 n.a. 0 n.a. n.a. n.a. 0 0 n.a.

184 Chapter E. Experimental results O EtOH 2 CO Others H 2 H 2 CO 10 H 4 C Biomass 8 H 3 C 6 H 3 C 6 H 2 C 4 H 2 C 4 CH a wt% mg/g Biomass Exp. #Liq.121 Char122 123 n.a. g Gas124 n.a. mg/g 125 n.a. Loss 126n.a. 0127n.a. 01280 1.3 1691290 1.5130 760 1.6 294 794 29131 1.6 270 338132 1.5 22 593 177 228 101 27133 1.5 161 289 37134 1.5 158 456 22 20135 13 65 1.4 165 135136 0 47 1.7 137 9 400 171137 1198 1.7 192 3138 0 0 1892 1.8 510 8 0139 4678 2.0 506 6 0 7 0140 17 3 0 1.8 468 102141 13 0 1.8 468 0 0 -9142 40 120 0 97 1.9 420 110143 0 10 0 143 0 18 1.9 194 0 106 182144 2.3 294 2 15 158145 0 17 46 0 14 2.0 232 0 133146 20 33 0 163 2.1 8 0 0 5 217147 31 0 10 1.8 14 216 1 264148 27 55 2 12 2.4 0 424 0 992 1391149 21 0 2.4 0 0 4 9 1386150 0 23 0 7 10 2.4 0 22 461 0 12 1259 0 0 107 24 2.3 3 33 420 0 1627 0 18 21 2.4 0 390 23 132 0 32 0 0 0 31 0 31 333 325 0 170 0 0 0 24 0 89 0 0 36 196 4 39 213 0 26 3 15 0 0 248 204 0 3 18 412 232 0 1 15 15 13 0 20 2407 2 0 15 0 0 0 18 42 0 39 10 3 21 18 0 650 0 131 0 260 0 0 0 5 11 7 0 30 60 349 0n.a. 0 34 0 13 9 0 31 26 0 11 52 128 0 0 34 0 26 176 11 0n.a. 0 0n.a. n.a. 202 141 10 5 0 24 32 0n.a. 0 175 0 0 0n.a. 0 00 n.a. n.a. 0 0 6 0 n.a. 0 0 0n.a. 0 n.a. 0 14 2 n.a. 0 0n.a. 0 0 0n.a. n.a. 0 0 60 5 0n.a. 0 00 0 68 0n.a. 0n.a. 0 0 19 410 69 0 000 0 6462 0 28 84 101 00 7401 0 38 8143 5 0 0 9800 8003 6053 0 21 42 99 13 39 136 6375 0 0 164 0 0 12 1367 1265 1302 7085 6052 0 17n.a. 00 44 0 0 0 20 0 1236n.a. 00 0 2 0 0n.a. 00 0 0 0 0 0n.a. 00n.a. 00 3 0 3 8 129 0n.a. 00 0 165n.a. 00 83n.a. 00 0 187 0 237n.a. n.a. n.a. 2316 0 0 n.a. n.a. n.a. n.a. 0 0 3 0 n.a. 0 n.a. n.a. n.a. 0 n.a. 0 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.

185 E.1. Tables of experimental results O EtOH 2 CO Others H 2 H 2 CO 10 H 4 C Biomass . 8 mg H 3 C 6 H 3 C 6 H 2 C 4 H 2 C 4 CH a wt% mg/g Biomass For experiments Exp. 54 and 55 without input of biomass yields are given in Percentage of total input. For Exp. 103-125 amount of converted catechol into not detectable substances a b Exp. #Liq.151 Char152153 1.6 g Gas154 1.4 mg/g 155 Loss 1.8156 18 1.9 259 519.9157 1.9 2653 335.8158 2081 1.8 339.0159 1.6 28 318.7160 22 1.3 93 13 285.2161 129 1.9 279.4 21 1.7 86 16 253.8 19 149 47 1.7 20 324.1 206 241.6 4 24 502 30 28 3 676 6 41 0 660 16 8 17 451 1 21 15 3 21 14 7 29 13 33 4 0 3 21 31 4 31 0 8 22 41 200 5 30 30 0 37 0 185 7 0 462 0 201 0 1 1496 0 0 16 0 55 0 14 0 2 1957 0 0 19 26 0 0 38 3 0 2 9 0 3 2 58n.a. 52 0 17 77 44 84 0 137 306n.a. 33 388 0 288 0n.a. 0 0 0 0n.a. 28 0 0 n.a. 29 0 0 11 31 0 n.a. 0 0 13 n.a. 0 n.a. n.a. 0 n.a. 0 0 n.a. n.a. n.a. 0 0 n.a. n.a. n.a. 0 n.a. 0 n.a. n.a. n.a. n.a. n.a. n.a. n.a.

186 Chapter E. Experimental results a hexO a 4-Me- 4-Et- Cyclo- Cyclo- a CatOH a Biomass 4-Me- 4-Et- Syr a Phe Phe Phe Phe Gua Gua CatOH CatOH hexOH 2-Me- 4-Me- 2-Et- 4-Et- Gua a Table E.3: Experimental data: Results of phenolic and other compounds quantiﬁed via GC - FID . # mg/g Exp. Phe 12345 16 27 18 29 0 110 0 111 0 112 0 1 0 13 0 1 1 14 3 0 0 n.a. 15 4 0 0 n.a. 16 1 0 1 n.a. n.a. 4117 1 0 0 0 n.a. n.a. 0018 1 0 0 n.a. n.a. 0019 1 0 0 n.a. n.a. 0020 0 0 0 n.a. n.a. 6221 1 1 0n.a. n.a. 5222 1 1 n.a. 0n.a. n.a. 6123 1 0 n.a. n.a. 0n.a. 52n.a. 24 1 0 n.a. n.a. 0n.a. 62n.a. 82025 1 0 n.a. 52n.a. n.a. 0n.a. 26 1 0 n.a. n.a. 62 n.a. 0n.a. 51027 4 0 n.a. n.a. 52 n.a. 15 0n.a. 93028 2 0 n.a. n.a. 51 n.a. 0 n.a. 51029 2 0 n.a. n.a. 72n.a. 16 0 n.a. 30 4 0 n.a. 51n.a. n.a. 0 n.a. 620n.a. 4 1 n.a. n.a. 41 0 5 n.a. 410n.a. 4 0 n.a. n.a. 52n.a. 420 0 n.a. 3 0 n.a. n.a. 62n.a. 32016 0 3 n.a. 3 3 n.a. n.a. 10n.a. 410 0 n.a. 2 n.a. n.a. 10n.a. 520 0 0 n.a. n.d. 1 n.a. n.a. 53n.a. 520 0 n.a. 2 n.a. n.a. 63n.a. 410 0 5 0 n.a. n.d. 1 n.a. n.a. 32n.a. 310n.d. n.d. n.a. n.d. 1 n.a. n.a. 11n.a. 410 n.a. n.d. 1 n.a. n.a. 53n.a. 520n.d. n.d. n.a. 1 n.a. n.d. n.a. 21n.a. n.d. 0 n.a. n.d. n.a. n.a. 42n.a. n.d. n.a. n.d. n.a. n.d. n.a. 52n.a. 221n.d. n.a. n.a. 62n.a. 211 n.d. n.d. n.d. n.a. 16n.a. 52n.a. 111n.d. n.d. n.a. 18n.d. n.a. 000n.d. n.a. n.d. n.d. n.a. 211n.d. n.a. n.d. n.a. 110n.d. n.a. n.d. 3n.a. 211 n.d. n.a. n.d. 4n.a. 211n.d. n.a. n.d. n.a. 421n.d. n.d. n.a. 311 n.d. 0 n.d. 0 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

187 E.1. Tables of experimental results a hexO a 4-Me- 4-Et- Cyclo- Cyclo- a CatOH a Biomass 4-Me- 4-Et- Syr a Phe Phe Phe Phe Gua Gua CatOH CatOH hexOH 2-Me- 4-Me- 2-Et- 4-Et- Gua a # mg/g Exp. Phe 3132333435 336 337 138 1 039 1 229 040 73 041n.d. 51 042 1 39 0n.d. 43 1 30 n.d. n.d. 44 1 n.a. 29 n.d. 45 0 n.a. n.d. 21 n.a. n.d. 46 0 n.a. n.d. 62n.a. 40 n.d. 47n.a. n.d. n.a. 42n.a. n.d. n.a. 48 1n.a. n.d. 62n.a. n.a. n.d. 49 1n.d. n.a. 51 n.a. n.a. n.d. 50 1n.d. 41n.a. n.a. n.a. 51 1 n.d. n.d. n.a. n.a. 0 n.a. n.d. 52 1 n.d. n.a. n.a. 0 n.a. 53 1n.d. n.a. n.a. 310 n.d. n.a. 0 n.a. 54 1n.a. 129 n.a. 210 n.d. n.a. 0 n.a. 55 1 0310n.a. n.d. n.a. 0n.a. n.a. 65 n.a. 56 1 0n.a. 410 1n.a. n.d. 26 57 1 1 n.a. n.a. 310 0n.a. 40 58 0 1 n.a. n.d. 0 0n.a. 59 0 1 n.a. n.d. 69 0 160 3 1 n.d. n.a. n.a. 0 131 3 1n.a. 0 2 0 14 n.a. 3 1n.a. 0 3 0 41n.a. 3 1 0 1 4 2 n.a. 3 0 10 2n.d. 5 2 0 0 1 3 n.d. 5 0 00 0 1 4n.d. 4 0 00 2n.d. 0 4 0n.d. 4 20 1 2n.d. 0 4 0n.d. 3 2 0 2n.d. 0 3 2 3 0 0 2n.d. 2 2 175 0 3 0 0 2 3n.d. 2 1 0 3 138 0 5 0 1 0 0 2 6 0 0 75 9 2 40 2 0 6 2 0 17 9 3 1 10 0 6 53 0 602 3 0 10 0 4 1 3 39 0 315 3 1 9 3 17 0 5 0 1 123 82 1 1 6 0 1 45 0 1 0 0 2 55 1 2 0 0 4 0 2 47 0 0 4 0 1 40 1 2 0 5 0 1 1 7 0 1 1 8 0 2 1 0 0 3 1 1 2 4 0 1 1 4 0n.d. 1 2 0 4n.d. 2 2 4n.d. 0 2 5n.d. n.d. 0 1 5n.d. n.d. 0 0 n.d. 4 n.d. 0 1 n.d. n.d. 0 0 4n.d. n.d. 0 4n.d. n.d. 0 n.d. n.d. 0 5n.d. n.a. 4n.d. 7n.a. n.d. 7n.a. 5 0n.a. 0 0 0 0 0 0 0 0 0

188 Chapter E. Experimental results a hexO a 4-Me- 4-Et- Cyclo- Cyclo- a CatOH a Biomass 4-Me- 4-Et- Syr a Phe Phe Phe Phe Gua Gua CatOH CatOH hexOH 2-Me- 4-Me- 2-Et- 4-Et- Gua a # mg/g Exp. Phe 3132333435 336 337 138 1 039 1 229 040 73 041n.d. 51 042 1 39 0n.d. 43 1 30 n.d. n.d. 44 1 n.a. 29 n.d. 45 0 n.a. n.d. 21 n.a. n.d. 46 0 n.a. n.d. 62n.a. 40 n.d. 47n.a. n.d. n.a. 42n.a. n.d. n.a. 48 1n.a. n.d. 62n.a. n.a. n.d. 49 1n.d. n.a. 51 n.a. n.a. n.d. 50 1n.d. 41n.a. n.a. n.a. 51 1 n.d. n.d. n.a. n.a. 0 n.a. n.d. 52 1n.d. n.a. n.a. 0 n.a. 53 1n.d. n.a. n.a. 310 n.d. n.a. 0 n.a. 54 1n.a. 129 n.a. 210 n.d. n.a. 0n.a. 55 1 0n.a. 310 n.d. n.a. 0n.a. n.a. 65 n.a. 56 1 0n.a. 410 1n.a. n.d. 26 57 1 1n.a. n.a. 310 0n.a. 40 58 0 1n.a. n.d. 0 0n.a. 59 0 1n.a. n.d. 69 0 160 3 1 n.d. n.a. n.a. 0 131 3 1n.a. 0 2 0 14n.a. 3 1n.a. 0 3 0 41n.a. 3 1 0 1 4 2 n.a. 3 0 10 2n.d. 5 2 0 0 1 3 n.d. 5 0 00 0 1 4n.d. 4 0 00 2n.d. 0 4 0n.d. 4 20 1 2n.d. 0 4 0n.d. 3 2 0 2n.d. 0 3 2 3 0 0 2n.d. 2 2 175 0 3 0 0 2 3n.d. 2 1 0 3 138 0 5 0 1 0 0 2 6 0 0 75 9 2 40 2 0 6 2 0 17 9 3 1 10 0 6 53 0 602 3 0 10 0 4 1 3 39 0 315 3 1 9 3 17 0 5 0 1 123 82 1 1 6 0 1 45 0 1 0 0 2 55 1 2 0 0 4 0 2 47 0 0 4 0 1 40 1 2 0 5 0 1 1 7 0 1 1 8 0 2 1 0 0 3 1 1 2 4 0 1 1 4 0n.d. 1 2 0 4n.d. 2 2 4n.d. 0 2 5n.d. n.d. 0 1 5n.d. n.d. 0 0 n.d. 4 n.d. 0 1 n.d. n.d. 0 0 4n.d. n.d. 0 4n.d. n.d. 0 n.d. n.d. 0 5n.d. n.a. 4n.d. 7 n.a. n.d. 7n.a. 5 0n.a. 0 0 0 0 0 0 0 0 0

189 E.1. Tables of experimental results a hexO a 4-Me- 4-Et- Cyclo- Cyclo- a CatOH a Biomass 4-Me- 4-Et- Syr a Phe Phe Phe Phe Gua Gua CatOH CatOH hexOH 2-Me- 4-Me- 2-Et- 4-Et- Gua a Exp. Phe 6162636465 366 367 368 3 269 2 070 3 071 2 072 2 2 073 0 2 074 0 2 075 0 2 2 076 0 1 1 077 0 2 2 078 1 1 2 9 079 1 16 0 0 080 1 14 0 2 081 0 14 0 0 1 0 682 0 12 0 0 0 683 1 15 0 0 0 784 1 13 0 0 0 6 085 1 11 0 0 5 0 586 1 0 0 5 0 0 687 1 0 0 5 0 0 588 1 0 0 14 0 4 0 089 0 0 13 0 1 4 0 090 1 0 14 0 3 4 0 0 0 8 1 0 15 0 5 4 0 0 6 1 0 10 0 4 0 0 7 0 1 0 13 0 4 0 0 7 0 1 0 13 0 5 0 0 5 4 5 1 0 0 5 0 0 6 4 1 0 0 0 3 0 0 5 5 1 0 0 0 2 0 0 1 1 0 0 0 5 0 0 4 0 4 1 0 1 0 5 0 0 3 3 1 0 0 0 6 0 0 0 3 0 0 0 0 6 0 0 0 1 0 0 6 0 0 0 4 1 5 9 0 0 3 0 0 2 2 1 7 0 0 1 0 0 2 0 1 7 0 3 0 0 3n.d. 1 4 0 7 0 0 3n.d. 0 1 8 0 0 8 0 0 5 n.d. 1 0 5 0 6n.d. 0 9n.d. 1 1 3 0 14 5n.d. 0n.d. 1 2 0 4n.d. 0 3 0 n.d. 1 2 0 0n.d. 0 2n.d. 1 2 1 0n.d. 0 4 2 1 1 0 4n.d. 0 4 n.d. 1 1 0n.d. 0 6n.d. 0 1 0 1 9n.d. 1 1 0 12n.d. 0n.d. 1 1 0 16n.d. 1 0 n.d. 1n.d. 1 2n.d. 2 1 n.d. 1 5n.d. 3 1 n.d. 4n.d. 9 1 6n.d. 7 n.d. 1 12n.d. n.d. 1n.d. 1 12 n.d. 1 1 n.d. 1 n.d. 3n.d. 2 n.d. 2 4n.d. n.d. 1 n.d. 5n.d. 1 n.d. n.d. 1 n.d. n.d. n.d. 0 1 n.d. n.d. 2 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. # mg/g

190 Chapter E. Experimental results a hexO a 4-Me- 4-Et- Cyclo- Cyclo- a CatOH a Biomass 4-Me- 4-Et- Syr a Phe Phe Phe Phe Gua Gua CatOH CatOH hexOH 2-Me- 4-Me- 2-Et- 4-Et- Gua a # mg/g Exp. Phe 9192939495 196 197 198 0 099 1 0100 1 0101 1 0102 1 1 0103 1 1 1 0104 1 2 0105 0 0 2 0106 45 0 0 0107 93 1 0 0 155108 1 0 0 0109 57 1 0 0 220 0 0110 0 0 209 0 0 0111 0 0 2 208 0 1112 0 1 341 0 0 1113 0 0 0 0 510 0 0114 0 0 3 552 0 0 0 1115 0 7 288 0 0 0116 0 0 0 5 481 0 0 0117 0 0 0 4 634 0 0 1118 0 0 0 3 202 0 2119 0 0 0 0 3 136 0 0 0 1 1120 0 0 254 0 0 2 0 0 1 0 0 0 343 0 0 0 0 0 0 0 0 412 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 19 0 0 0 0 0 0 0 16 0 0 0 0 0 0 0 0 0 0 14 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 9 0 0 0 7 0 0 0 0 0 0 8 0 0 9 0 0 0 0 0 12 7 0 0 0 0 0 0 15 0 0 1 0 0 0 0 17 0 3 0 0 18 2 0 0 0 0 0 2 0 18 3 0 0 4 0 2 0 0 16 778 0 0 0n.d. 6 0 0 802 0 1 0 0 565n.d. 8 0 0 8 0 1 0n.d. 0 0 560 9 0n.d. 1 0 413 0 1 0 n.d. 9 0 0n.d. 0 240 0 2 0n.d. 0n.d. 0 396 0 0 3 0n.d. 3 0 334n.d. 0n.d. 0 0 3 0 256n.d. 0 0n.d. 0 3 0 0 264 n.d. 0n.d. 0n.d. n.d. 0 0 0 0 0 59 n.d. n.d. 0 201 0 0n.d. n.d. 0 0 110 0n.d. 0 8 0 436 0n.d. 0 0 11 0n.d. 0 59 0 0 0 71 1 0 0 3 73 1 23 207 0 0 0 0 3 0 0 0 3 0 0 2 28 0 0 0 0 0 0 120 3 0 97 1 0 37 0 0 1 69 0 8 141 21 316 86 95 7 7 27 125 70 5

191 E.1. Tables of experimental results a hexO a 4-Me- 4-Et- Cyclo- Cyclo- a CatOH a Biomass 4-Me- 4-Et- Syr a Phe Phe Phe Phe Gua Gua CatOH CatOH hexOH 2-Me- 4-Me- 2-Et- 4-Et- Gua a # mg/g Exp. Phe 121122123 138124 265125 270126 139127 0128 93 0129 0 0130 0 0 0131 0 0 0132 2 0133 0 3 0 0134 0 3 0135 0 0 3 0136 0 4 0 0137 0 0 1 0 138 0 0 0 1 0 n.a. 139 0 0 1 0 n.a. 0140 0 0 2 0 n.a. n.a. 33 0141 0 2 0 0 n.a. n.a. 54 0142 0 2 0 n.a. n.a. 86 0143 0 0 3 0 n.a. n.a. 86144 0 0 0 4 0 n.a. n.a. 97145 0 0 1 0 n.a. n.a. 74146 0 0n.a. 0 0 0n.a. n.a. 44147 0 n.a. 2 0 0n.a. n.a. 30 0n.a. 0 148 0 n.a. 0 0 0n.a. n.a. 32n.a. 4 149 0 0 n.a. 1 0 0n.a. n.a. 73n.a. 4 150 0 0 n.a. 1 0 n.a. n.a. 94n.a. n.a. 4 0 0 0 n.a. 2 0 n.a. n.a. n.a. n.a. 4 0 0 n.a. 2 0n.a. n.a. 44n.a. n.a. 5 0 418n.a. 3 0n.a. 0n.a. 83n.a. n.a. n.a. 4 012 n.a. 1n.a. 31 69 n.a. n.a. n.a. 4 112n.a. 0 0n.a. 00 54 n.a. n.a. n.a. 2 1n.a. 0 0 n.a. n.a. n.a. n.a. 3 0 1 28 0 1 5 0n.a. n.a. n.a. n.a. 6 1 5 0n.a. n.a. n.a. 0 0 n.a. n.a. 1n.a. 0n.a. n.a. n.a. 0 0n.a. n.a. 2n.a. 0n.a. n.a. n.a. n.a. 0 0 n.a. n.a. 0n.a. 0 0 n.a. 5n.a. n.a. 0n.a. n.a. 0n.a. 4n.a. n.a. 0 0n.a. 0n.a. 2n.a. 0 0 264n.a. 26 n.a. 210 0 14 0n.a. 37 0 n.a. 1 17 0 0n.a. 38 39 n.a. 0 0 0 91 n.a. n.a. 67 0n.a. 0 10 n.a. n.a. 0 0n.a. 41 0n.a. n.a. 2 0 62 0 103 n.a. 0n.a. 0 0n.a. 1n.a. 0 0 45 n.a. 0 2 0n.a. 0n.a. 0n.a. 0 n.a. 10 0n.a. 0n.a. 0 1n.a. 0 19n.a. 0n.a. n.a. 0 1 15 3 1 15 0 12 8 14 1 5 1 9 8 0 7 3 n.d. 8 0 2 n.d. 5 n.d. 2 n.d. 2 n.d. n.d. n.d. 2 n.d. n.d. 2 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

192 Chapter E. Experimental results a hexO a 4-Me- 4-Et- Cyclo- Cyclo- a CatOH a Biomass 4-Me- 4-Et- Syr a Phe Phe Phe Phe Gua Gua CatOH CatOH hexOH 2-Me- 4-Me- 2-Et- 4-Et- Gua a Phe =phenol, Gua=guaiacol, Syr=syringol, CatOH=catechol, CyclohexOH=cyclohexanol, CyclohexO=cyclohexanone. a # mg/g Exp. Phe 151152153154 0155 0156 0157 0158 0 0159 0 0160 0 0161 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0n.d. 0 0 0n.a. 0 0 0n.d. 0 0 0n.a. 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

193 E.2. Energy and mass ﬂow

E.2 Energy and mass ﬂow

In this section the energy content of the commonly applied biomass and of gaseous and solid products are presented (see Table E.4). The HHV was calculated consulting the results from CHNS analysis (see Section 2.3.4) employing the method by IGT [Tal81]. Additionally the mass and energy balance for experiment Exp. 66 is presented in Table E.5 scaled to 1 t/h lignin.

Table E.4: Heating value and elemental composition of input and output components for solvolysis of wheat-straw lignin in ethanol and formic acid in a CSTR at 673 K, 25 MPa and 44 min mean residence time. Component Results from elementary analysis HHV C H N S O Ash % % % % % % kJ/kg wheat-straw-lignin a 63.8 6 1.5 0.3 27 1.4 24955 Spruce-lignina 60.5 6.7 0.6 0.1 30.1 2 24305 Ethanol 52.2 13 0 0 34.8 0 26865 Formic Acid 26.1 4.3 0 0 69.6 0 4561

H2 0 100 0 0 0 0 120900 CO 42.9 0 0 0 57.1 0 10107

CH4 75 25 0 0 0 0 50144 C2H4 85.7 14.3 0 0 0 0 47246 C2H6 80 20 0 0 0 0 47590 Others 52.2 13 0 0 34.8 0 26087 Chara 76.7 5.2 0 0 16.7 n.a. 29882 liquid productb 37.7 12.3 0 0 50 n.a. 20413 a heating values estimated from elemental analysis with the method by IGT [Tal81]. b Heating value from energy balance.

The enthalpy of each input and output stream H˙ i was calculated by multiplying the HHV with the mass ﬂow m˙ i. Exceptionally the HHV of the oil fraction, that is the liquid product without ethanol and phenolics, could not be measured. However, the enthalpy of the overall liquid fraction (ethanol, phenolics and oil) was determined. In addition, the enthalpy of the sum of all products H˙ Sum could be calculated by adding the enthalpy of all three product phases. The enthalpy of the oil fraction was thus determined

194 Chapter E. Experimental results

Table E.5: Mass and energy balance for experiment Exp. 66 scaled to 1 t/h lignin.

Input m˙ i H˙ i Output m˙ i H˙ i t/h MJ/h t/h MJ/h EtOH 9.48 254680 Ethanol 8.13 218395 Formic Acid 1.22 5564 Phenolics 0.07 1907 Lignin 1.00 24955 Gas 1.14 17119 Char 0.02 518 Oila 2.34 12536 Sum 11.70 285200 Sum 11.70 250476

∆HR -34724 a not measured, see Equation E.1 by diﬀerence, according to Equation E.1 X H˙ Oil = H˙ Sum,Output − H˙ i. (E.1) i6=Oil

The reaction enthalpy ∆HR was calculated according to Equation E.2.

∆HR = H˙ Sum,Output − H˙ Sum,Input. (E.2) Lignin degradation in ethanol and formic acid has a reaction enthalpy of -34724 MJ/h and is thus exothermic. The results obtained from the energy balance were used for the feasibility study.

195 E.2. Energy and mass ﬂow

196 Appendix F

Cost oﬀers

In this Appendix important cost offers are presented which were used for the calculations within the feasibility study. The first offer for a plunger pump was obtained from Schäfer & Urbach. The specifications of the pump are given in the offer-page. The second offer from ICIS is a cost information about the market prices of phenol in the first half of the year 2012. The market price of benzene is also included.

197

Phenol (Europe)

Editor Julia Meehan, [email protected]

CONTRACT PRICES Click for Price History Price Range US CTS/LB FD NWE JUN EUR/TONNE +73.00 1585.00-1625.00 +73.00 91.10-93.40

SPOT PRICES Click for Price History Price Range Four weeks ago US CTS/LB FOB RDAM T2 EXP USD/TONNE +5 1165-1190 +20 1200-1210 52.84-53.98 FD NWE EUR/TONNE n/c 1320-1350 n/c 1250-1260 75.87-77.59 EX-WORKS RUSSIA Rb/TONNE -500 49500-55500 -500 53500-59000

NOTE: for full details on the criteria ICIS pricing uses in making these price assessments visit www.icispricing.com and click on “methodology”.

Phenol producers and consumers reduce operating rates Spot phenol activity is very thin, since there is no demand. The majority of phenol producers and consumers have Producers and consumers are unable to suggest even a reduced operating rates because demand has slowed in notional range because there is simply no interest for spot Europe and for exports to Asia. volumes. However, producers have said that there is no need to move spot prices below the benzene contract price plus Reducing production ahead of the peak summer months of €300/tonne. July and August is common and most sources said this week they would not be reducing output more than is normal for the The third-quarter adder over the benzene contract price time of year. However, some buyers said their own customers The third-quarter fee discussions are ongoing. further downstream are considering extending outages for economic reasons. Consumers continue to push for reduction and in some cases a decrease of €5/tonne has been achieved. In general, the phenol market in Europe is described as relatively balanced, but cutbacks in operating rates by Exports of phenol to Asia producers and consumers of around 20%, has prevented an As has been the case for many months now, outside long-term imbalance in the market. fixed contract volumes, exports to Asia have been non-existent and European spot prices have simply been tracking spot price Demand downstream has clearly slowed, more so for some developments in Asia. derivatives than others. Those affected most are the epoxy resins and polycarbonates (PC) markets because of This week in Asia, spot prices in Asia moved up by $5- competitively prices imports. 20/tonne of the CFR China Main Port price range.

In the nylon intermediate markets, the lack of European Russia exports moving to Asia and margin pressure has meant that Phenol prices in Russia moved down because of increasing some makers of nylon and nylon intermediates are considering stocks and poor demand. Local phenol prices were expected taking plants off line if market fundamentals do not improve in to remain stable to soft in coming weeks. Russian July. domestic numbers include 18% VAT.

Feedstocks July contract and benzene The European propylene contract price for July settled at The phenol market has been keeping a close eye on the spot €935/tonne FD NWE, down by a record €170/tonne from June, benzene market this week in the hope that trading activity and because of weak upstream naphtha and ongoing soft supply- the direction of the spot market might give some clue to the and-demand fundamentals. This is the lowest contract price possible outcome of the July benzene contract price. since October 2010. The market viewed the settlement as a necessary adjustment given the challenges facing the Even up until the close of business on Friday, downstream European industry. Sources are hopeful that the new price will markets reliant on the benzene contract price were still in the be the bottom of the market and will encourage some renewed dark about the direction of the market. interest for derivatives.

However, ahead of business close the benzene contract price for July settled at €953/tonne FOB NWE, down by €68/tonne. Production update PKN Orlen started its planned maintenance on phenol and Spot phenol acetone production at its facility in Plock, Poland, on 22 June. The plant is due to restart on 1 August.

Copyright violation is a serious offence. Any distribution or forwarding of information which is not expressly permitted by your subscription agreement is a copyright violation. ICIS pricing will be using software to monitor unauthorised electronic redistribution of reports. Copyright 2012 Reed Business Information Limited. ICIS pricing is a member of the Reed Elsevier plc group.

Phenol (Europe)

Page 2 of 2

This week on ICIS www.icis.com Domo will start a planned outage at its facility in Leuna, 28/06/2012 18:28 Europe polyamide chain producers mull Germany, on 1 July for two weeks. shuts if margins stay low 28/06/2012 16 PKN starts planned phenol/acetone France’s Novapex will undertake a regular summer shutdown maintenance shutdown during the first half of August. 29/06/2012 08:1 6China’s May cyclohexanone exports fall 74% month on month Polimeri typically has a biennial shutdown in Mantova, Italy, and this is scheduled for next year.

($1 = €0.80) (€1 = Rb41.22)

FEEDSTOCK PRICES (CONTRACT) BENZENE Click for Price History Price USD/GAL FOB NWE JUL EUR/TONNE -68.00 953.00-953.00 -68.00 4.02-4.02

ICIS pricing accepts no liability for commercial decisions based on the content of this report.

For information about multiple Copyright violation is a serious offence. Customer Support Centre subscriptions and licences to this Any distribution or forwarding of information +44 20 8652 3335 or toll free from information product, or for permission to which is not expressly permitted by your US/Canada:+1 888 525 3255 photocopy or redistribute individual reports, subscription agreement is a copyright ICIS pricing:www.icispricing.com please call the relevant office: violation. ICIS pricing will be using software ICIS News:www.icis.com/news London:+44 20 8652 3335, to monitor unauthorised electronic ICIS website:www.icis.com [email protected] redistribution of reports. Copyright 2012 Houston:+1 713 525 2600, Reed Business Information Limited. ICIS [email protected] pricing is a member of the Reed Elsevier Singapore:+65 6789 8828, plc group. [email protected]